[Date Prev][Date Next][Thread Prev][Thread Next]   [Date Index] [Thread Index] [Author Index] 

Re: simulation vs. mimesis

Sometimes, I get really dense. Like right now. I hope you'll forgive me.

A model airplane, hell, look at what I just said! "model" airplane. In every sense of the word, my model airplanes are real airplanes. They actually do what all airplanes do (we're not talking about those plastic ones that break if you toss them), except, maybe, carry people or serious cargo. They still interact with their environment in every sense of the ways in which the big behemoths do. They just don't do so in precisely the same way, scale effects again.

So, my model airplanes are not models at all. They are airplanes. But, at the same time, they *can* be models if airplanes if their intent is to *simulate* the behaviors of a larger vehicle they were built to, um, model. That word "model" again.

Now, let's really talk about models of airplanes. (That word "model" again). Those are purely inferential constructs, sometimes crafted in Fortran. I wrote all mine in Forth, and, later in a lisp dialect (Scheme) that I wrote in Forth. this would be the so-called "empty simulation." It's in the context of this particular use of the word "model" that I would agree with the statements that have been made earlier. There, the simulation turns out to be a graph with colored lines that show this or that parameter plotted against other parameters. But, in the context of the earlier (see below) use of the word "model" (that of a flying object that generally has all the important appendages and features of a "real" airplane), that I remain, perhaps, confused. That would be because the specific phrase used is *model airplane*, not *airplane model*. Model airplanes, to me, have *never* been those I wrote as software; the software variants were always "airplane models" or just "models".

So, now I turn to some new words.
<quote>
A model airplane either "simulates" or "models" the appearance of an airplane (without including much, if any, of the underlying causal entailment relations of the airplane in its own inferential entailments...
</quote>

I am interpreting that passage to mean this:

Givens:
*Model* (flight type, not computer type) airplane has its own modeling relation, one with the actual model and its causal entailments mapped to one that has inferenential relations, those one imagines to see in a bunch of differential equations.

*Real* (flight type, carries people)airplane has its own modeling relation, one with actual airplane and its causal entailments mapped to one that has inferential relations, those one imagines to see in a bunch of differential equations.

Inference:
The inferential relations of *Model* can't commute with the causal relations of *Real*.

I guess I'm not ready to accept that, or I don't understand it at all.
I thought I already suggested that, indeed, a model airplane (flight type) does, indeed, include the necessary observable and causal entailments, which, in my judgement, implies that the necessary inferential entailments are there too, albeit adjusted for scale effects, and I have yet to see why those won't map or commute with the same or similar structures for "real" airplanes. What am I missing?


Jack

Judith Rosen wrote:

*A model airplane either "simulates" or "models" the appearance of an airplane (without including much, if any, of the underlying causal entailment relations of the airplane in its own inferential entailments). If the inferential entailment relations between the model's external observables are not in any commutative relation with the causal entailment relations which exist between the airplane's external observables, then it's not a "model" in the sense we have been discussing; it is a "simulation". The modeling relation doesn't commute. Which is my point: the modeling relation is A RELATION. So, the relation, as specified, is what determines whether something is a "model" or an empty "simulation"... How we define these things is context based (context perhaps being: What is the model supposed to do?) So the definition of "a simulation" is sometimes radically different from that of "a model" and other times they are synonymous. If the relations for "external observables" commute, but we want to model something about flight of an airplane, then the external-observables-model isn't a useful model. Again, the difference is based on contextual constraints.*
** *Judith*

    ----- Original Message -----
    *From:* Tim Gwinn <mailto:***>
    *To:* *** <mailto:***>
    *Sent:* Friday, December 31, 2004 1:14 PM
    *Subject:* Re: [ROSEN] simulation vs. mimesis

Judith,
JR:*What I'm saying is that a simulation is still a model when the
entailment structures of the external behaviors/observables are what
are intended to be modeled (my example of the movie set and the storm).*
What I am saying is that if the entailment structures are in
congruence, then it is a model, not a simulation. How is a "model
airplane" a /simulation/?? As I quoted, 'simulation' requires an
extraneous machine on which the simulation runs, and which does not
decode to something on the natural system side.
Regards,
Tim


        -----Original Message-----
        *From:* ROSEN Forum [mailto:*** Behalf
        Of *Judith Rosen
        *Sent:* Friday, December 31, 2004 11:03 AM
        *To:* ***
        *Subject:* Re: simulation vs. mimesis

*Tim, there is more to entailment congruence between system and
models than that-- it sounds like you are saying that a model
has to model the whole entailment structure to be considered a
model! This is not the case. How do you think humans have been
able to get as far as we have, scientifically??? *
** *Yes, a modeling relation has the requirement of congruence
between causal entailment in the system and inferential
entailment in the model, but no model can include the
system's entire entailment. All models are incomplete
representations. This is simply the nature of the beast as far
as human capacity to do science goes. The only big problem with
it arises when humans forget that our models are incomplete
representations of natural systems and we start making decisions
based on predictions learned from our models for applications in
the real systems. This is exactly what we are doing with genetic
engineering of crops to be tolerant of herbicides, for example.*
** *What I'm saying is that a simulation is still a model when the
entailment structures of the external behaviors/observables are
what are intended to be modeled (my example of the movie set and
the storm). A model airplane, for example, has almost none of
the underlying causal entailment of a real airplane. It's use as
a model is limited to only a few practical applications: Like
how to design an airport hanger to house the airplane if it is
that design of airplane and the model is "in scale". What does
"in scale" mean? It means that entailment relations in external
observables are preserved, regardless of the different size. If
the model is in scale, then those relations will commute with
the relations present in the actual system (the plane). *
** *A flight simulator is also a model, but not of the entailment
relations of the external observables. It models other
entailment relations. Do you see?*
** *Judith*
****