Re: Modeling vs. Simulation [from "fundamental problems in physics"]

[Tim Gwinn <***>]

JohnK,

See interposed.

Regards,

Tim

-----Original Message-----
From: ROSEN Forum [mailto:***]On Behalf Of John Kineman
Sent: Monday, December 27, 2004 6:36 PM
To: ***
Subject: Re: Modeling vs. Simulation [from "fundamental problems in physics"]


Tim & Boris,

TG: "Boris nicely summarized "simulation" in his recent post: "let us understand that, through simulation, inferential entailments, which play the role of efficient causes, become material causes. And that makes a clear distinction between simulation and the modeling relation, because the modeling relation "respects" inferential structures"."

Yes, this makes a lot of sense to me - an excellent distinction if applied deeply, I think. It also agrees with Tim's comment that simulations are mechanistic.

However, it still does not provide a way to distinguish when one would call any given surrogate system an instance of a model or an instance of a simulation. It speaks more to the process of generating these surrogacies, the modeling relation being richly complex and on-going, whereas a simulation is intended as an approximation without further efforts to improve its congruency. The point I was makeing is that in the lab, we see only the snapshot of the modeling process, hence a single instance of a model, and that is what I was questioning if we could objectively distinguish it from a simulation.

 

 

TG: Depends on what you mean by "objectively". If science is the epistemological attempt to model the natural world, then I think this "objectively" criteria is an attempt to apply a meta-criteria to the modelling relation: how do we independently (that is, outside of any modelling relation) know that what we are modelling is accurately represented by our models? But this would be metaphysics, since any scientific evidence would itself be in the form of a modelling relation. As mere mortals, we are limited to our measurements, predictions and creative skills as modelers. 

 

 

The fact that the model taken as part of a modeling relation is thus part of an ongoing process of commutation, whereas a simulation is not generally considered to be (although we do try to improve simulations ???), is perhaps as good an answer as I can get.

 

TG: Time would not be a good criteria, because a model might be a valid model, but one which is valid only for a short timeframe; and a simulation might simulate over very long periods of time. The distinction is in the model maintaining a congruence of entailment structures with the object system, whereas the simulation has no such concern.  

Some fine points:

The requirement of "commutation" is also not so clear to me, as anything that resembles nature at all can be said to commute with some aspect of it.

 

TG: This is why mathematics has such value in science. It comes to the scientific community already with a general consensus of fixed definitions and fixed rules of operations. (By this I do not assert that mathematics is Platonic or anything like that, I only mean that societally, mathematics enjoys a general consensus which extends to the scientific community.) This allows us to spend efforts on creating encodings/decodings to such models, rather than having to spend time trying to interpret or agree on what the elements of the model mean, such as in art, for example, where interpreting the elements and relations in the art 'model' is always "in the eye of the beholder".

 

 

The criteria of reflecting "real" entailment structures seems unapproachable because we never know if we built a model or simulation on similar principles as the natural system or not. That is a matter for experimentation.

 

TG: This seems to imply that there are two distinct things: the modelling relation on the one hand, and experimentation on the other hand. In the Rosennean epistemology, these are not distinct. Experimentation without a model is pointless. A commuting modelling relation has both encoding and decoding: measurement and prediction. A modelling relation thus inherently embodies the empirical, the experimental, as well as the theoretical.

 

I would thus find it hard to distinquish simulacra from model in a practical sense - it would have to be a judgement of how "real" we think the model is based.

 

TG: This is true for all observations in science. How "real" is the measurement of mass, for example? How do we "objectively" verify that? The subjectivity of all of science thrusts its head prominently forward here. We cannot ever be meta to our observational limits. This goes for all aspects of the natural world -- entailment relations are no more or less epistemologically suspect than any other. 

 

For example, we would have called Ptolemy's model of the solar system a model as long as we believed nature was based on perfect circles. When our belief about that changed, it became a classic example of a simulacra (of which any finite instance would then be a simulation).

 

TG: Yes, all models are subject to revision or replacement upon further knowledge. That has always been a part of doing science.

The important point I come away with is that the modeling relation "respects inferential structures" as real in their own right, and implicit in any natural system such that they are then capable of disrupting any precise specification of that system.

 

TG: I don't understand this last point. I'm not sure what you mean, but I would not say that inferential structures are "implicit" in any natural system.  Also, can you elaborate on what you mean by the "capable of disrupting any precise specification of that system" remark?

Thanks!!<>

JK