This demo was originally designed to
model clustering of ethnicities by
neighborhood, given even a minor
preference for neighbors of their own
kind.

Here the actors are amphibians.

The original model, as sociology, I'm
told has been refuted as too simple, but
it still makes a fun demo.

The frogntoad.slogo demo has a subtle
but gross counting error introduced by
me, the translator.

Use the likeuntolike.slogo version
instead, it is a lot more fun anyway.

Since the demos have no internal
documentation, this will have to do.

What is happening is some breeds of
amphibians are being scattered at random
over the patch screen.

Some land on top of one another, so
their first goal is to find an empty
space into which to unpile.

Their remaining goal is to move away if
neighbors not of their kind outnumber
neighbors of their kind too much.

The resulting emergent behavior is that
like clusters with like in visually
distinguishable patches.

The demo as posted took five hours to
achieve a 33% clustering ratio.

You might like it that way, or you might
reduce the patch area and the number of
each breed of amphibian tenfold or so to
see the results faster.

A mix of 45% empty patches and 55%
occupied patches is one setup that works
pretty well.

This demo code originated for me with
Brett Smith, brett@att.com.  He got it
out of a book.

Now you can also try wanderlust.slogo,
which adds to the original idea the
concept that sometime folks just move to
move, maybe to add a bedroom, regardless
of matching like for like surrounding
neighbors.

Probably their search for a compatible
neighborhood should start at a random
empty spot on the screen, but for now
they just take a step and then seek from
there for a compatible neighborhood
again.

These demos and this note are kept at:

http://www.well.com/user/xanthain/public/code/StarLogo/FrogsAndToads/

Cheers!

xanthian.

Kent Paul Dolan, xanthian@well.com