Posts tagged as:

System Dynamics

Today’s New York Times features an article, “We Have Met the enemy and He Is PowerPoint” that became the day’s most emailed piece. It features a complex diagram designed to present the relationships that shape counterinsurgency dynamics. This isn’t even a new story: as my colleague John Sterman reminds me, it was in the news last year. John argues that the story offers a good illustration of the poverty of the mental models most people hold and suggests checking out Stephen Colbert’s quite funny piece about this, Afghandyland.

I have to say I got many messages about the NYT article myself!  Why the interest?  In my circle, it’s partly because the author conflates the complexity of a causal loop diagram designed to capture a full spectrum of COIN, or counterinsurgency, policies and dynamics with the use of powerpoint. I’m happy to critique powerpoint–in fact, it’s been a bit of a theme for me these past five years–but I also don’t want to unfairly blame the medium for problems in presenting dynamic complexity.

First, on presentations as the method of sharing ideas and work. We are seeing how important powerpoint is everywhere, even in my students’ class projects this semester where some of our project clients want only powerpoints. The form is, clearly, here to stay, at least for the present–Tufte’s earnest injunctions notwithstanding (check out his monograph with the enticing subtitle “Pitching Out Corrupts Within”). For some useful advice, feel free to check out some great blog posts (such as this one from Lifehacker),  aptly-titled books like “Death by Powerpoint,” and my favorite, Garr Reynolds. And get better at using powerpoint. Try new things, experiment, iterate, weed out what doesn’t work.  Run your presentations by a “murder board” of peers or colleagues. In almost every case, you’ll need to cut slides and words, simplify diagrams, and leave off things.

The result will be slide decks that are even less self-explanatory when they stand on their own [click to continue…]


We’ve had a long debate within the field about the pros and cons of using standard generic structures to represent model elements. I tend to  be wary of their use, thinking that it’s important to approach a given modeling project with as few preconceptions as possible. But if social science and psychology are to be at all useful, they must present us with phenomena and explanations that apply across different cases. This suggests that there are indeed “generic structures” in the world. Let’s look at a few starting points. First, we look at the more heuristic approaches commonly called systems thinking, then at computer-simulation approaches.

Peter Senge’s book presented a set of “system archetypes” that boil down some commonly-found structures into causal loop diagrams and connect them heuristically to behavior over time. Archetypes are

images of common systemic situations. Each of these patterns occurs in a wide variety of domains, from ecology to economics to manufacturing; each offers its own strategic insights, and gives people a better picture of how the forces of the system may trap them.

System dynamics researchers have published descriptions of about a dozen archetypes. They include “Limits to Growth,” in which a seemingly boundless growth pattern runs up against unexpected limiting forces. (Total quality campaigns, for example, run up against institutional disappointment after the “low-hanging fruit” is picked.) In another archetype, “Shifting the Burden,” a more immediately inviting, short-term solution to a problem weakens the system’s ability to develop a more fundamental, but slower, approach.


To learn more about the idea in general as well as the specifics of the archetypes [click to continue…]


So, one way to think about limits to growth writ large is to use the example of given societies and argue that the same holds more generally. The MIT System Dynamics Group followed a different approach in the early 70s when it undertook a study of the “global problematique” at the request of a group of industrialists and thinkers called the Club of Rome. The MIT researchers built a simple model of the world (the World 3 version of the model was most fully documented in 1974’s Dynamics of Growth in a Finite World). I won’t go over more details here, but you can find materials on line and most of you will already have (and I hope already explored) the World model that comes with Vensim. If you haven’t taken a look at the model, please do.

In 1993 a 20-year update was published, called Beyond the Limits, and 2004 saw the 30-year update ( )

There have been other global models—here, too, you can find much information online (see, for instance, ), with interest in simulating the “global problematique” seeming to go through varied phases over the decades. Recently, of course, climate modeling is a hugely important aspect of global modeling.

It’s easy to critique simple global models like the World models, but the question that the field keeps coming back to is: does it represent core dynamic processes that drive large-scale change? And can we glean useful insight from them? To consider this last question, I want to lay out some ideas about the forces that shape global limits to growth.

What are the global limits to growth?

Let me share some resources to get you thinking. First, One report to read in this area is Graham Turner’s “A comparison of The Limits to Growth with 30 years of reality,” Global Environmental Change, Volume 18, Issue 3, August, Pages 397-411. (a version available at ). In another recent assessment, Hall and Day revisit the limits to growth after peak oil in a 2009 American Scientist piece ( )

I mentioned in class a Wall Street Journal piece that takes on the topic: as well as some useful resources from the New Scientist ) As with much of my material in this area, thanks to Tom Fiddaman and his metasd blog for great resources.

Dennis Meadows reflected last year on the proximity of global limits to growth His colleague Jorgen Randers also reviews evidence in a 2008 paper in Futures entitled Global Collapse—Fact or Fiction? (found online here:

And an excellent big-picture view is presented in Herman Daly’s 2005 Scientific American paper, Economics in a Full World Daly, Herman E. “Economics in a Full World.” 293, no. 3 (September 2005): 100-107. Daly argues that

the global economy is now so large that society can no longer safely pretend it operates within a limitless ecosystem. Developing an economy that can be sustained within the finite biosphere requires new ways of thinking.

If all this leaves you in need of cheering up, check out the Happy Planet Index.


We talked in class about the general phenomena of encountering limits or constraints to growth. Any reinforcing or growth process eventually slows down. Often, there are saturation effects. For instance, in the Bass diffusion model, as the stock of adopters grows larger and larger, there’s a corresponding drop in the stock of potential adopters. Eventually, the decline in potential adopters slows down the rate of adoption, just by the simple “physics” of the adoption process.

Other effects can slow growth. Competitors may enter a market, thereby slowing the growth of the focal firm. Consumers may demand more for their money as they become more sophisticated, and be less prone to adopt. Once they’ve made an initial purchase, consumers may become less enamored with a given product and fail to replace it when it wears out.

Limits to growth in populations

In the general case of a population (people on Easter Island or yeast in a petri dish), limits to growth occur as the level of population grows relative to the carrying capacity of the environment. Rather than adoption being the driver of growth in such cases, it’s reproduction. Net growth rates (reproduction-deaths, per time period) fall as population grows relative to carrying capacity, driven by crowding and competition for scarce resources.

In many ways, it’s a classic model; see Section 4.2 in Sterman’s Business Dynamics. Differential-equation models have been published in several literatures (e.g. ecological economics and physics) . In system dynamics, an interesting study presents and examines a model of a society that had its own cultural practices that limited population growth—helping to explain how one society avoided total collapse. Kampmann’s paper aims to replicate and explore an earlier system dynamics model, so it’s worth looking at both for the model, theory, and evidence, as well as for the methodology and approach. See Kampmann, Christian. 1991 “Replication and revision of a classic system dynamics model: Critique of population control mechanisms in a primitive agricultural society.” System Dynamics Review, 7(2): 159-198. And since we’re in this brief sidebar of discussing model critiques, let me also point you to Liz Keating’s paper Issues to Consider While Developing a System Dynamics Model –huge thanks to her and Tom Fiddaman for sharing.

Modeling limits to growth for a population requires you to consider a few things carefully. One, you want to think carefully about time constants and delays. Why would populations overshoot their carrying capacities? Why isn’t feedback instantaneous? Remember to think about the “physics” of the processes you are attempting to capture.

You’ll also need to represent the carrying capacity of the environment. Clearly, you’ve got to get the boundaries right! For instance, what’s the model boundary for the case of Easter Island (also known as Rapa Nui)? Pretty clear that it’s the island itself, right? The story, after all, is that the population grew to—and beyond—the capacity of the island’s ecosystem to support people.

But here’s an interesting question. Original settlers landed on the island in boats. Why didn’t they simply leave on boats once things began to look dire for them? From what we can tell, because the population denuded the island of trees, the material to build boats was simply unavailable. And the population pressures caused internal strife, thereby reducing islanders’ abilities to collaborate on system-level solutions to their challenges. Many of you have read Jared Diamond’s Collapse: How Societies Choose to Fail or Succeed (2005, Viking); if not, you can get the main ideas in 18 minutes via this Ted talk: .

Now, it wouldn’t be science if there weren’t a controversy, and when it comes to Easter Island, one debate is about the effect of the first arrival of Europeans. The argument is that the inhabitants of the island did survive deforestation and evolved into a more egalitarian society, for a while, only to be hit by disease from European seafarers. For more on this rival explanation, see: Easter Island’s Controversial Collapse: More To The Story Than Deforestation? ScienceDaily (Feb. 18, 2009)

So, there’s debate about Easter Island, but, thanks to Diamond and others, there are plenty of other Limits to Growth cases to consider. Check out this NASA story about the fall of the Maya. The title says: “They did it to themselves.” (anyone wondering why NASA is interested in this issue?).


Here’s my current course, a new take on an old classic, Applications of System Dynamics. This semester’s version focuses on Global Challenges.

these folks are old hands at applying system dynamics