Re: [Swarm Modelling] Successful predictions are not enough

[Jason Alexander]

Hi Cal

> 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)

This is absolutely right and (to my mind) crucially important.  So far 
I've been ignoring the fact that  explanations are often, though not 
always, answers to why-questions and, as such, usually pitched at a 
particular level (and with a certain set of terms and concepts in mind) 
and any explanation that is to count as an "acceptable" or 
"satisfactory" explanation needs to be addressed at the same level.  My 
favourite example is the (hypothetical) answer to the why-question "Why 
did my computer crash?"  Now, if the only criterion of a satisfactory 
explanation is that it be a causal explanation, then one could answer 
the question of why my computer crashed by handing me a complete binary 
state description of my computer one minute before it crashed.  That, 
plus knowledge of how my computer changes state, counts as a complete 
causal explanation of the phenomenon.  But I'd be really ticked off if 
a MSCE responded to my question of why my computer crashed by dropping 
a 500-page book of 0s and 1s on my desk.  The level of explanation 
matters.

Cheers,

Jason
--
Dr. J. McKenzie Alexander
Department of Philosophy, Logic and Scientific Method
London School of Economics and Political Science
Houghton Street, London WC2A 2AE