Because Mike complains about the proceeding of building models after
making experiments, that can explain the results, I cite H.R.Erwin:
"Modellers who seek predictive results are indeed asking the wrong
question much of the time, and instead should be seeking insight into
their system." ("The Dynamics of Peer Polities", 1990, named
CambridgeText.ps + CambridgeFigs.ps at anonymous ftp-site
`mason1.gmu.edu' under /herwin.)
As I understand, the insight is, that you know what parts your model
is built of (attractors, saddle-points, limit-circles... or:
information, replication, mutation, selection... or: whatever). If you
find a real pendant for all the parts, you can simulate what will
happen, when a new attractor/selection/... occurs, two
saddle-points/species/... merge and so on. That doesn't mean you can
make any exact predictions about the future development of your
system.
Joachim (joaccigh at w250zrz.zrz.tu-berlin.de)
