Re: validation from data is also insufficient

[cal]

alex wrote:
...
On another note, if there are errors that are more likely to be made
by all human modelers (be it from quantum brain wiring or such), then
surely all fitting models at least represent human understanding of
the topic better than only one would? 
...
Several models are built until there's more data to kill the unfit ones.

cal:
yes, this certainly should be true - my real point is that more than the usual
``curve fitting'' is required to use data to evaluate models, even though the
argument becomes very subjective - this is the source of much argument, even
in mathematics (Kronecker on Cantor's cardinal number theory that allowed and
required computations with infinite sets, ``That is not mathematics.  That is
theology.'')

alex wrote:

RS>ralf wrote:
RS> Should reread my Horgan

Surely you don't mean John Horgan?  Now there's my idea of an ontological
hell: doomed to re-reading the `End of Science', again and again and again and
again...

cal:
uh, yeah 8-) - i really don't like horgan's book at all, because it is an
announcement of defeat - as i wrote long ago,

       a failure of imagination is not a proof of impossibility

(though i would love it if somebody can find another source for this quote,
since i am not sure i invented it entirely by myself) in any case, when you've
proven that something can't be done, it is time to see which of your
assumptions you can change or leave out to make progress (there are many new
mathematical developments that arose this way, from self-referential sets to
context-aware logics)

more later,
cal



                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================