Words to replace "model"

CSGnet Archive CSG1997
1

[Martin Taylor 970225 1550]

Bill Powers (970224.1135 MST)]

Martin, I have no recollection at all of this:

>What we agreed was to try to avoid the use of the
>term "model" in both cases, and replace it with terms like simulacrum
>or internal replica, on the one hand, and mould or inverse on the other.

I said I was going to use the term "simulation".

You are right. I should not have said "agreed", since it was my agreement
with what I thought you had said you intended, not a mutual agreement on
terminology:

That is all I agreed to; I left the uses of "model" up to you, since you
seemed to insist on using it to mean a lot of different things.

Maybe you remember [Martin Taylor 970131 12:30]:

+I have been using the term "model" to indicate any representation from
+which an appropriately constructed analyst could recover the form of
+what was modelled, not in the narrower sense of something that looks
+like what is modelled.
+...
+But I stand by what I said, in the important sense that the environment
+determines the range of structures and parameter values that yield functioning
+control structures, and the fact that a particular control structure
+functions can be used to delimit the range of environment within which
+it resides. Whether we use "model" or "simulation" or "replica" or
+"inverse analogue" is important for communication. Of that list, I think
+"inverse analogue" may be the most suited to a HPCT structure, as a mould
+fits its casting, whereas "replica" may best suit the system that produces
+output that stabilized the pseudo-CEV in a replica world, in the hope that
+the real CEV in the outer world will thereby be stabilized.

ยทยทยท

+
+So perhaps we should entirely eschew obfuscation, and not use "model" at
+all. It will be hard for me, and, as you have said, you also slip.

If you like
you can use it to mean simulacrum

(= simulation)

or replica

(= simulation)

or mould or inverse

(= inverse analogue)

or
perceptual function or output function or reference signal or anything you
like; it no longer has any technical meaning, by my understanding of the
agreement.

Those things are not of the same nature as a model, in the sense of that
from which a copy of an original can be reconstructed with more or less
fidelity. Any of them could, in principle, contain or be the result of
using a model, but they could not be models in either the sense of
simulation or the sense of inverse analogue.

Simulation, however, does. Not "simulacrum" or "replica" but
"simulation." That's the only term to which I respectfully lay claim.

I wasn't actually trying to introduce new terms. I was trying to point out
that we had (I had on my side, and you had on your side) agreed not to use
the word "model" where "simulation" or "inverse analogue" was intended.
Simulacrum and replica seemed to me also to make the point, as did "mould"
and "inverse", under appropriate circumstances.

Rick was using a "simulation", not an "inverse analogue" in his sinusoidal
tracking experiment that he chooses to call a demonstration of "two-level
control."

Martin

2

[From Bill Powers (970225.0640 MST)]

Martin Taylor 970225 1550]--

If you like
you can use it to mean simulacrum

(= simulation)

or replica

(= simulation)

or mould or inverse

(= inverse analogue)

or
perceptual function or output function or reference signal or anything you
like; it no longer has any technical meaning, by my understanding of the
agreement.

Those things are not of the same nature as a model, in the sense of that
from which a copy of an original can be reconstructed with more or less
fidelity. Any of them could, in principle, contain or be the result of
using a model, but they could not be models in either the sense of
simulation or the sense of inverse analogue.

I think you are confusing a similarity in _behavior_ with a similarity in
_organization._ A simulation shows no particular behavior. If I simulate a
lever with the equation y2 = k*y1, I am saying nothing about how either y2
or y1 behave. If you tell me how y1 is behaving, I can predict from the
simulation how y2 will behave, but without that information the simulation
is only a representation of an organization. The same applies to the control
system simulation. A control system simulation shows no particular behavior
until you specify the behavior of the independent variables, the disturbance
and the reference signal. Then we can predict the behavior of the perceptual
signal, error signal, and output of the system.

This is why, when you speak of a _signal_ as a model of something, you can't
possibly mean what I mean by a simulation. And a replica is certainly not a
simulation, except in the specific case where the _functions_ in the
simulation as well as their connections are the same as those in the real
system. A simulation subject to one set of input conditions may behave quite
differently from the real system subject to a different set of input
conditions, yet the simulation could still be quite correct. Conversely, you
could have a simulation that shows the same detailed behavior as the real
system under a given set of conditions, yet it could be incorrect, as
changing the conditions would show.

I hope this shows that the concept I call "simulation" is very different
from the meanings of the other terms you use.

Rick was using a "simulation", not an "inverse analogue" in his sinusoidal
tracking experiment that he chooses to call a demonstration of "two-level
control."

Not as you seemed to understand the term. You interpreted Rick's explanation
to mean that he simply supplied a sinusoidal reference signal, and called
this a "simulation." But that would not be a simulation, because it doesn't
explain where the sine-wave comes from.

I remember a bit more, now. If I remember correctly, the aim was to build a
control system using an oscillator as an output function, with error signals
that altered the amplitude and frequency of the oscillation. That would make
the system into a true simulation, since the organization that functioned as
an oscillator would produce no particular frequency or amplitude until the
error signals were specified. This would still be an interesting project,
but it was never finished.

Best,

Bill P.

3

[From Bruce Gregory (970226.1000 EST)]

Bill Powers (970225.0640 MST)

I think you are confusing a similarity in _behavior_ with a similarity in
_organization._ A simulation shows no particular behavior. If I simulate a
lever with the equation y2 = k*y1, I am saying nothing about how either y2
or y1 behave. If you tell me how y1 is behaving, I can predict from the
simulation how y2 will behave, but without that information the simulation
is only a representation of an organization. The same applies to the control
system simulation. A control system simulation shows no particular behavior
until you specify the behavior of the independent variables, the disturbance
and the reference signal. Then we can predict the behavior of the perceptual
signal, error signal, and output of the system.

I think of a simulation as an entity whose functioning is
intended to tell us something about the functioning of the
world. It is as if we were to say, "If the world works
according to these principles, we can expect to see the
following occur..." In addition to principles, you also need
initial conditions in order to "run" the simulation. This seems
completely consistent with your description, and I will assume it
is until you tell me otherwise...

Bruce Gregory