|
With regards to Robert Rosen's work: There has been a bit of
confusion over when, or if, a simulation is also a model-- and the
complementary issue of when, and why, it isn't.
As it has been pointed out, a model can be either good or
bad and in my father's view, good refers to how well the entailment structures
in the model correspond to the system being modeled. At this point, the word
"model" could be anything. A landscape painting is a model, for example. So
is a photograph. A computer simulation could be a type of model as well (weather
forecasting). Not all models are simple systems, either; a rat in medical
experiments is a model of a human. This goes into my father's discussions of
"surrogacy" and the aspect of whether one model is better than another in any
given situation is to enter into my father's extensive treatment of issues he
called "optimality" (on which he wrote an entire book; his first, entitled
"Optimality Principles"). However, we can state categorically that all
simulations are simple systems. This is a given because one of the tests for
complexity is that a complex system has at least one non-computable model.
I think it would be reasonable to say that if a simulation has any
entailment structures that are congruent to the system it simulates, then it is
technically "a model" of that system. However, there are times when it would be
inappropriate to call the activity of creating simulations "modeling". When the
intention in creating the simulation is to simply mimic epiphenomena of the
system in question, with no interest in the underlying causal organization, then
you have a "simulacrum" which is of extremely questionable scientific
value as a model. In that case, my father would make a distinction between the
simulation and a model.
There may also be the case where you are trying to create
realizations of a model or a set of models, and this may or may not be a
simulation. This goes into the discussion on analytic vs synthetic models.
However, even in this case, the realization is a form of model. Again; it might
be a lousy one. There is no value specified in the designation "model", other
than that at least one entailment structure is congruent to the system
originally modeled. (See chapter 6 in "Life, Itself" for Analytic and Synthetic
Models.)
Judith Rosen
Note: The case of creating a simulacrum in order to learn about
organisms is one my father felt was particularly unhelpful and often very
damaging. He said this is exactly what Descartes did, in effect. I recently did
some research into the work of Stephen Wolfram and his computer simulations of
life. It certainly looks like he's going to inadvertantly test this part of my
father's theory.
|