[From Bruce Abbott (960907.1015 EST)]
Bill Powers (950906.1125 MDT) --
One point that's been overlooked in the God of Rules discussion is
whether any _properties_ of the objects are used in the description.
. . .
This is the problem with proposing rules that aren't grounded in the
properties of the actual system. While a rule may appear to work, it may
be nothing more than a restatement of the observations, a tautology, or,
because of the way it happens to be formulated, it may actually imply
properties that don't or can't exist. A wrongly-stated rule can send
everybody off onto wild goose chases that have no hope of ultimate
success. And this is true even if the rule appears to conform to the
observations.
What you say about descriptive models is right: they are useful when
they are the best we can do. But descriptive models are just that; they
are simply descriptions of what is observed. They may sometimes be cast
in terms of explanations, but the explanations are always just tight
little circles.
I agree for the most part--certainly one should prefer a mechanistic model
over a purely descriptive one, for the reasons you (and I) have stated. But
it seems to me that some models that appear at first glance to be purely
descriptive are actually mechanistic, though lacking in certain details.
The physicists' rule assumes that the particles interact in a particular
way, without stating the mechanism of that interaction. Newton did the same
thing when he posited the inverse-square law of gravitational attraction: he
imposed a rule which he applied to any particle having mass. That rule
turns out to be enormously useful, even though Newton had no firm idea how
forces like gravitation could act at a distance as specified by his rule.
Fortunately for Newton, his rule seems to hold rather generally. Our
physicists' rule undoubtedly does not, because the unspecified mechanism
through which their rule appears will produce other apparent relationships
when conditions vary. A truly general explanation of the sorts of behavior
which emerge from the physicists' rule will require, as you so rightly point
out, a model that incorporates the essential physical properties of the
"particles" involved.
Regards,
Bruce
[From Rick Marken (950907.0845)]
Bruce Abbott (960907.1015 EST) --
it seems to me that some models that appear at first glance to be purely
descriptive are actually mechanistic, though lacking in certain details.
A truly general explanation of the sorts of behavior which emerge from the
physicists' rule will require, as you so rightly point out, a model that
incorporates the essential physical properties of the "particles" involved.
Let's try not drop the ball, here. The problem with the physicists' model is
NOT that it is descriptive rather than mechanistic. Nor is the problem that
it is lacking in certain details. The problem with the physicists' model is
that it is a CAUSE-EFFECT model of CONTROL.
Best
Rick
[Martin Taylor 950907 12:30]
Rick Marken (950907.0845)
Bruce Abbott (960907.1015 EST)
A truly general explanation of the sorts of behavior which emerge from the
physicists' rule will require, as you so rightly point out, a model that
incorporates the essential physical properties of the "particles" involved.
Let's try not drop the ball, here. The problem with the physicists' model is
NOT that it is descriptive rather than mechanistic. Nor is the problem that
it is lacking in certain details. The problem with the physicists' model is
that it is a CAUSE-EFFECT model of CONTROL.
Sorry, puzzlement here (as with each time you've said this). I thought
that the physicists were describing the behaviour of groups of mutually
influencing particles. If there's control anywhere, isn't that
control something within each individual particle? I don't remember
Tom's descriptions making any mention of what the physicists said was
happening within a particle.
Or are you now saying that crowds are control systems, as opposed to
the results of interactions amng many control systems?
Martin
[From Bruce Abbott (950907.1245 EST)]
Rick Marken (950907.0845) --
Bruce Abbott (960907.1015 EST)
it seems to me that some models that appear at first glance to be purely
descriptive are actually mechanistic, though lacking in certain details.
A truly general explanation of the sorts of behavior which emerge from the
physicists' rule will require, as you so rightly point out, a model that
incorporates the essential physical properties of the "particles" involved.
Let's try not drop the ball, here. The problem with the physicists' model is
NOT that it is descriptive rather than mechanistic. Nor is the problem that
it is lacking in certain details. The problem with the physicists' model is
that it is a CAUSE-EFFECT model of CONTROL.
Rick, what I'm doing here is trying to formulate (as much for my own benefit
as for anyone else's) how various models differ. As I've thought about the
problem more, it has seemed to me that the physicists' model is not purely
descriptive after all, but does present a mechanism (read causal structure)
however vaguely specified, in the same way that Newton's model of planetary
motion differs from Kepler's purely descriptive laws. Kepler's laws
describe the planetary motions but all you can do with them is plug in
starting conditions and solve for position, velocity, and acceleration at
time t. The laws just summarize the observations. Newton deduces Kepler's
laws from certain "rules" he applies to the centers of mass. The same rules
lead to correct descriptions of all sorts of other motions, such as pendulum
swings.
In the same way, local rules applied to the particles in the physicists'
model lead to certain aggregrate behavior. The rule itself does not merely
describe the aggregate behavior, as a purely descriptive model would. Thus
the physicist's model functions more like Newton's than Kepler's.
You wish to call attention to another problem with the physicists' model,
and rightly so. To the extent that living organisms "follow the rule," they
will behave as the model predicts. But only a generative model which
recognizes that the "particles" involved are control systems can explain why
the physicists' rule appears to hold (when and if it does). The physicists'
model is cause-effect, but it is not a model of control at all. At the
level of abstraction at which the physicists' model applies (if it applies
at all), it simply says that under these conditions, this will happen. As
such is it neither compatible nor incompatible with the more fundamental
control-based model.
Regards,
Bruce
[From Tom Bournon (950908.1509)]
From Rick Marken (950907.0845)]
Bruce Abbott (960907.1015 EST) --
it seems to me that some models that appear at first glance to be purely
descriptive are actually mechanistic, though lacking in certain details.
A truly general explanation of the sorts of behavior which emerge from
the physicists' rule will require, as you so rightly point out, a model
that incorporates the essential physical properties of the "particles"
involved.
Let's try not drop the ball, here. The problem with the physicists' model
is NOT that it is descriptive rather than mechanistic. Nor is the problem
that it is lacking in certain details. The problem with the physicists'
model is that it is a CAUSE-EFFECT model of CONTROL.
I agree, Rick, that the physicists' model is one of lineal cause and
effect. However, I still think it is worth emphasizing that it is a
cause-effect model that is intended to create a re-creation of the original
observation: objects that "tend" to move in the same direction as their
neighbors. It is a cause-effect model for the outward appearances of a
phenomenon. Like Bruce said in some of his earlier posts on this tread, it
is a model at a higher level of abstraction -- for sure. 
Bruce, I'm about to put together a reply to your latest to me, about a day
and a half ago. We are agreeing so much that I can no longer rely on heat
and flame to inspire my replies. 
Later,
Tom
