Re: Nonsimulable models

["Jeff Pridaux" <***>]

Excellent post Tim.

In Anticipatory Systems Rosen said something like the "reactive paradigm"
is universal in the sense that it can be used to describe the behavior of
any natural system...(curve fitting).   But the reactive paradigm ignores
all the causal entailment.

Once you have the data, you can always "react to it" after the fact and
form a "formal system" that can re-generate the data.

It is probably true that the behavior of a natural system containing
impredicative causal loops can be simulated with a simulation (without any
impredicative causal loops) but as Tim says, the simulation wouldn't
actually be a model since it didn't reflect the causal structure present in
the natural system.

The problem is that most scientists prefer to just consider that the
natural system doesn't have the extra entailment (to make it match the
entailment of the simulation) and then effectively consider the simulation
(or mechanism) to be a model.