Re: Models with Meaning LO1285

Richard Karash (
Thu, 18 May 1995 23:24:33 -0400 (EDT)

Replying to LO1166 --

Mike and all, sorry to be slow in replying, but the admin of learning-org
is taking enough time that some evenings I just don't have time to
contribute to the conversations myself. It's not lack of interest...

On Fri, 12 May 1995, Michael McMaster wrote:

> And Rick responds,
> > I can make the analogy more easily in the physical sciences, so I'll
> > start there. Kepler's laws, although a huge breakthrough at the time, were
> > only descriptive. He said "Planets move in an ellipse." But, he couldn't
> > say why this should be the case.
> Bateson's development of "an explanatory principle" may be useful
> here. In fact "gravity" was one of the terms he used to make the
> point - although I think "instinct" was the better known one. In each
> case, the "explanation" stops the inquiry at that point. It's not
> whether or not the explanation is valid nor useful. It's just that,
> so far in history, the inquiry has stopped at the current meaning of
> the term. Fish return to their streams "by instinct" was the end of
> the conversation - until someone came up with a useful explanation
> that went beyond "instinct".

Gravity is a very interesting example. It's such a part of our
experience, we know it intuitively. It's so obvious. We *forget* about
explaining it, and it's pretty hard to explain.

> Rick says:
> >(Newton) provided the why and how; a structural explanation,
> > that was reasonably complete and satisfying. Operationally, it's good
> > enough to sent satellites into orbit, man to the moon, even the Voyager
> > probe to the outer planets.
> Here, I think, is the problem. "a reasonably complete and
> satisfying" explanation has as its standard something other than a
> pragmatic test. "instinct" was a reasonably complete and satisfying
> explanation. And probably still is for many people for many things.

Mike, I think it was and is pretty transparent that "instinct" is
not a satisfactory explanation, that when people refered to "instinct"
it actually meant, "We don't know how!"

> However, it "explains" nothing. More like, it explains it away. Its
> the second part of what Rick puts forward that is key for me. That
> is, _operationally_ its "good enough to send satellites into orbit."

I think that "does it help with x?" is another dimension. Many
descriptive models can help operationally; it all depends on what you're
trying to do. In the planetary example, the descriptive model is useful
for finding stars, knowing when the planets will be visible, etc. But,
not enough for space travel.

An example of a descriptive model that's very useful operationally -- As
a teenager, I was on a farm, intrigued by a steel cable connecting a
tractor to some equipment being towed. A farmer boy told me in a thick
drawl, "Don't stand near that cable! If it breaks, it'll kill you."
Operational, yes. But, I'm sure he didn't know why or how, or how to
define the limits of the danger zone.

> Now _I_ don't need another explanation. What I'll want is
> satisfaction that I have a "likely story" that is in fact related to
> the pragmatic effect of putting something into orbit.

How to get this satisfaction?

1) "face validity" -- does the model seem like a reasonably satisfactory
explaination? Does it make sense? Of course this can be misleading. This
is the notion of "reasonably satisfactory" explanation.

2) Does it explain the observed behavior? This is the usual scientific
method, making sure the theory is consistent with the data.


         Richard Karash ("Rick") |  <>
     Innovation Associates, Inc. |    email:
 3 Speen St, Framingham MA 01701 | Host for Learning-Org Mailing List
 (508) 879-8301 and fax 626-2205 |     <>