Wayne wants to "understand the real power of models to demonstrate that they
generate less error than a disciplined mind." He then goes on to say, "Why
shouldn't verbal hypotheses be just as explicit as those inserted into a
computer program?" This is why I have long advocated the use of fuzzy logic
in ecological modelling. The disciplined mind also uses models, but because
it doesn't need to boil these down to computer code the models can be much
more flexible and comprehensive. Fuzzy logic comes much closer to the way
that we think than standard computer programs.
For example, those who have worked on harmful algal blooms know that a
period of warm calm sunny weather is likely to produce toxic dinoflagellate
blooms. What could be a simpler model? OK, now program that into a computer.
Define the temperature range for "warm". Do we define "calm" in terms of
mean wind speed or the peak of gusts? Does "sunny" involve photoperiod? If
by ecological models you mean only those that sit in a computer, they aren't
very good. If you mean models that incorporate as much expert knowledge as
we have available, even if it is not always quantifiable, we can do much
better.
Bill Silvert
----- Original Message -----
From: "Wayne Tyson" <[email protected]>
To: <[email protected]>
Sent: Thursday, January 22, 2009 10:40 PM
Subject: Re: [ECOLOG-L] ecological modeling
I would like to see a specific case example of ". . . a model that is
capable of accurate predictions over some limited time scale" so I can
begin to understand the real power of models to demonstrate that they
generate less error than a disciplined mind. I am not implying that I do
not "believe in" models; I mean exactly what I said. An explanation of the
input, processing, and output also would help. In other words, can someone
"demystify" ecological models for me in the spirit of advancing this
particular discussion?
John, what "context" are you talking about? What, exactly, makes you
cringe?
I like the idea of being explicit, with or without models. Why shouldn't
verbal hypotheses be just as explicit as those inserted into a computer
program?
WT
