You are currently browsing the category archive for the ‘System Dynamics’ category.

In computer programming, we often avoid introducing global variable because it is bug-prone and hard to trace. Local variable is the first choice.

In designing systems, to control the oscillation of the system, we try to introduce quick local [negative] feedback loops to avoid major [negative] feedback loop with delays to be triggered, which often bring unfavourable consequences.

I was amazed when I discovered the similarity of these two ideas today. Do they belong to some more general principle?


First of all, in this Year of Tiger, 虎 (you) wants to be a millionaire?

To me, I have no interest in being a millionaire. I just hope my income (if not sufficent in the time of difficulties) and wealth can cope with my consumption in my lifetime:
Stock and flow map of income, Wealth and consumption

So besides trying to increase my income and wealth to meet my consumption rate, I also control my consumption by spending only on the true needs.

That’s why I can be happy.

Major fishery collapsed, and I did not care about – because I was rich enough to have the remaining fish;
There were more and more wildfire, and I did not care about – because I am not living near the forest;
Some countries faced water scarcity, and I did not care about – because in my place, rainfall is more than enough;
Sea level rise is inundating river delta, island nations and coast, and I did not worry about – because I can always move to my summer house in the highland;
The high food price prolonged hunger, and I did not worry about – because my country is self-sufficient in grains;
Oil is getting expensive, and I did not worry about – because my country exports oil;
Then the world war began and epidemic outbreak – and there were no one left to help me.

The boiling frog slows to react because the changes are gradually. Something changes gradually is not salient, the skill we developed during the evolution process to select things to pay more attention. Unfortunately, overshoot symptoms are often distant in space and time, therefore often less salient.

For example, sea level rise is almost unnoticed because it is very slow. But if it is consistently rising, one day we will find ourselves need to move to higher place after more and more severe flooding occured. The increase in flood risk is the indirect outcome of sea level rise.

It is well-documented that people performed poorly in simple stock and flow task, especially when the flows information are displayed as Behavior Over Time graph, the difference between flow and stock becomes less salient. Event is more salient than process/bahaviour, which is still more salient than the underlying structure, needless to say the prevailing mindset (mental model, paradigm) which bring up the structure.

I denote such fallacy which only focus on the salient things and ignore the important but less salient things as saliency trap.

… because it takes time for people to adjust their perception, once the initial perception has been set.

The adjustment time can be very long, depends on the person. That’s why it is the best to give people a good first impression. There is a Chinese idiom for this – 先入为主.

A simple bathtub Interactive Learning Environment (ILE). Can you control it so that water fill the bathtub fully but no overflow? What is the Behaviour Over Time Graph (BOTG) of the inflow to do it?

Jay W. Forrester’s “Learning Through System Dynamics as Preparation for the 21st Century” may give you a clue:
the latest version as mentioned by Bill Harris.

A very good introductory book of systems thinking by Donella Meadows after I peeked some pages from
Google Books.

A series of Interactive Learning Environments (ILE) that illustrate the prevalence of dynamic complexity in everyday life:
At Any Rate

Download the models and try them out!

Some lectures about the exponential growth. There are 8 parts in this series.

Read the rest of this entry »

System Dynamics (SD) can be seen as one kind of visual representation of nonlinear control (or nonlinear dynamics in Physics), especially for application on social problems (management, economics, politics, etc). Therefore it is no surprise to find that many System Dynamicists have natural sciences or engineering background. In fact, Jay W. Forrester himself was an engineer.

While System Dynamicists commonly use software such as Vensim, iThink (Stella) and Powersim, it is actually tool independent and can be implemented in many ways.

Here is an implementation of SD library using Modelica (open source competitor of MATLAB). While it is still quite basic and not customized for SD quick use, it is probably a good bridge for newcomers to learn SD if they come from engineering background. Its interface is similar to engineering software.

For a quick glance, check out its slides and paper.

If you want to try out:

Latest Tweets


Blog Stats

  • 6,150 hits
Creative Commons License
unless otherwise noted.
%d bloggers like this: