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

Re: simulation vs. model



Tim,
 
Define "simulation".
 
To simulate something is to... what?
 
In the example you cite, there is no argument over the fact that simulation is not a congruence between entailment structures. However, there is congruence over external observables. So... in the case where you are modeling a storm for some film you're directing (setting: "... it was a dark and stormy night..."), the external observables are those properties you need to model. In the case where you are trying to predict what the weather in some location is going to be like on next Thursday... you must model different aspects which correspond to entailment structures. My father's point is that human beings tend to think if they can simulate a system convincingly enough, that's the same as creating an exact replica... and human beings also tend to think that, because we can fool ourselves and each other with a convincing simulation, that systems which "seem" complex are really simple underneath.
 
Judith

 
----- Original Message -----
From: Tim Gwinn
To: ***
Sent: Thursday, December 30, 2004 11:23 PM
Subject: Re: [ROSEN] simulation vs. model

Judith,
 
I disagree entirely that simulations are also models, in the sense in which your father used those terms. He explicitly distinguishes between modeling and simulations in LI ch. 7. On p. 200, for example: "Simulation is thus not a congruence between inferential entailments."  A simulation is not a modelling relation -- it fails the requirement that the encoding/decoding bring the two entailment structures into congruence.
 
Regards,
Tim
 
 
-----Original Message-----
From: ROSEN Forum [mailto:***On Behalf Of Judith Rosen
Sent: Thursday, December 30, 2004 10:50 PM
To: ***
Subject: Re: simulation vs. model

I've said it before but it bears repeating: All simulations are also models. My father said many times that any system can be "simulated". That means there are aspects of any given complex system which can be reproduced with high "commutativity" with the original system. What he was pointing out was that just because some model gets external, observable aspects of the system right doesn't mean it gets all aspects right. In other words, don't forget that the simulation is only a model of the system, not a reproduction of an identical system.
 
Any "random" set we generate becomes a repeating pattern if you close the loop.  Randomness is a local, finite property and is in the eye of the beholder.
 
Judith