Gordon wrote
[delete]
> Jim does have a point. Mathematics does allow a great deal of flexibility
> for model building. While not all things are possible, there are lot of
> possibilities that incorrectly describe reality. A good illustration can
> be given in physics: What happens when Newton's inverse square law is
> replaced by an "inverse cube law" in the gravitational equation? Nothing
> that resembles reality. But it is no more than a mathematical model - and
> one can easily illustrate its effects on the computer. But it is just the
> wrong model. Nothing more.
>
> A "proof" of a good model is that it makes good predictions. The inverse
> square law certainly meets this test. At present, does any mathematical
> model of evolutionary processes make good predictions?
>
> Gordon Simons
>
A couple of years ago I modeled the entire National League season when
there were just two divisions. I started out with all the offensive
and defensive data for each team. But I didn't put in the results for
a single game--no information on who won, who lost. (I actually wanted
to prove the Cubs had really won their division, but that didn't work.)
Then I ran off season after season on my computer until the cumulative
average wasn't changing any more. Results: ten of the twelve teams
were in the correct positions for that year. The won-lost was only
slightly wrong for one of the others. The twelfth, the Dodgers, was
significantly higher in my model than they actually were.
I don't know what this proves, except to show I'm a better manager
than Tommy La Sorda.
Russ
e-mail: rmaatman@dordt.edu Home address:
Russell Maatman 401 Fifth Ave. SE
Dordt College Sioux Center, Iowa 51250
Sioux Center, Iowa 51250 Home phone: (712) 722-0421