[Swarm Modelling] Successful predictions are not enough

[Chris Landauer]

hihi, all -

i've been following this discussion of modeling with interest, and i want to
point out that good predictions are not always enough - ``good'' explanation
is also an important and useful criterion (even if it is hard to decide what
``good'' means in this context)

there is one very good example of this situation - Ptolemy's hypercycles were
MUCH more accurate predictors of planetary position than Copernicus' model for
many years - until Kepler and the notion that orbits might be ellipses instead
of circles, and until Newton had an explanation that guaranteed ellipses

this warning is especially important in a modeling application that relies on
emergent behavior - phenomena are easy to observe, hard to measure, and
explanations are hard to come by

more soon,
cal


Dr. Christopher Landauer
Aerospace Integration Science Center
The Aerospace Corporation, Mail Stop M6/214
P.O.Box 92957
Los Angeles, California 90009-2957, USA
e-mail: address@hidden, Phone: +1 (310) 336-1361