Complex Systems Summer School 2000
StarLogo Project: HomeWork Problem #2
One of the homework problems Ronnie gave us was to show that the unequal
distribution of wealth phenomenon can be seen in an easy StarLogo simulation.
The turtles start out with the same amount of money, $100 for example.
When they meet each other, one randomly chosen gives the other a dollar
if they have at least one to give. The point of the exercise is that
there occurs a considerable imbalance of wealth.
Java/StarLogo
In order to do this on my Linux system, I used the Java based version of
StarLogo. It was a somewhat painful process due to the combination
of the Linux JDK and the StarLogo system. Crashes were frequent and
the UI was flaky. It also proved frustrating for an experienced programmer
to use a system designed for non-programmers! The manual used an
incremental learning technique which was hard for me to use. I never
could figure out how to have a procedure return a value, for example.
Similarly I could not construct a string composed of values of variables
with interspersed text .. a "printf".
None the less, it was a LOT of fun! The system really is simple
and the simplicity makes the agent behavior all the more clear.
The Program
Because the StarLogo system is somewhat incomplete, I had to use a combination
of gnuplot and StarLogo: StarLogo would run the agent simulation,
and the result of a manual "show money" command in the turtle's command
window would be cut & paste into a file, which gnuplot would then graph.
Below is the startup configuration; the first with no steps taken by
the turtles, the second with just 11 steps. Note that they are all
initially all yellow and the second window shows half yellow, half blue.
This is due to using four colors for the turtles, depending on their wealth:
$0 -$50: Red, $50 -$100: Blue, $100 -$150: Yellow, above $150: Green.
Click for larger image
The associated graphs from gnuplot are here. Note the first is
simply a uniform $100 distribution. After 11 steps, there is some
slight difference of wealth.
The Results
After running for several thousand iterations, the disparity in wealth
is pronounced. The first snapshot is at 3600 iterations, and the
second at 6000.
In the first case, there are a few folks with $0, and some with as many
as around $280. In the second, there are a few more $0 folks, and
a few with as many as $325.
