Application design

[Sven Thommesen]

I have just created my first Swarm app, a modification to Heatbugs. I found
this to be surprisingly 
easy. Kudos to the design team! 

However, I can see where doing more complex things can get ... well, more
complex. So, before I
get in too heavy, a design question:

if I want to design an app that has say 5 types of agents all living on the
same space, and I want to
have probes (averagers etc) on each of the agent types as well as on
individual agents, which is 
the better (easier, more extensible) way to go:

a) one ObserverSwarm which owns one ModelSwarm which owns 5 lists of
different agents;
b) one Observer Swarm which owns 5 ModelSwarms (and in this case, how is
ownership of the
   underlying space done?)

I welcome any suggestions.

Sven Thommesen
UCLA Center for Computable Economics
Auburn University School of Human Sciences