Learning from experience LO6450

Fri, 5 Apr 96 15:47:10 -0600

I recently overheard these statistics in a coffee conversation about
learning. 70% of learning comes from new experiences, 20% of learning
comes from new people, 7% of learning comes from formal education, and 3%
goes for error. I recall that anything can be proven with statistics when
dealing with quantitatively illiterate and non-thinking person who lack
curiosity. Let s just think about those numbers AS IF they were true. We
hear #1-Some people have 30 years of experiences and others have 1 year
experience 30 times. #2-Lateral job changes are touted as good for
*growth*. #3-The job makes the person. These sayings all promote the
concept of learning from new experiences.

In his recent book, Mike Mc Master [The Intelligence Advantage:Organizing
for complexity, 1995] says: Our personal competence is measured by how
quickly and effortlessly we can produce results in the face of
interruptions, breakdowns, and problems in performance.[page 269] This
concept applies to organizations as well. A breakdown is "a declaration
of interruption to the flow of a system to focus attention on a need for
learning and on demands beyond the current ability of the system to
handle". [page 281] To transform our organizations, we must be willing to
welcome breakdowns as opportunities for learning or even to cause
breakdowns to create the opportunity for learning. A recurring breakdown
must be transformed into an occasion for learning. A smooth and
uninterrupted flow of production does not support (or promote)
organizational growth and development. For instance, JIT was intended "to
promote learning while producing, rather than for production itself."
[page 277].

In a recent OR article,["Learning from Mistakes: A Note on Just-In-Time
Systems", Operations Research, Volume 44, Number 1,
January-February,1996.] Ocana and Zemel, present an analysis of
Just-In-Time Systems that contrast the classical inventory management with
Just-In-Time (JIT) management principles. They partially reconcile the
two frameworks by interpreting JIT within the optimization paradigm. They
"claim that some of the operational characteristics of JIT, as opposed to
its rhetoric, can be obtained within the "optimization" framework,
provided that additional relevant factors, previously ignored are taken
explicitly into account."

Their model incorporates three elements:
. A standard dynamic optimization model of operations
which includes production, demands, uncertainties, and
. A standard model of induced learning which allows for
gradual improvement in the operational process.
. An observability constraint which couples the
learning and the optimization process. This new
constraint postulates that learning occurs through
experiencing shortages or "learning from mistakes".

This formalism allows the resolution of improvement into a standard
learning effect (present in all learning) and a JIT accelerated effect
induced by operating at a sub-optimal level. The JIT system benefits over
a standard learning system by "manufacturing" mistakes from which to
learn. The "lowering the lake to reveal the rocks" analogy expresses the
later idea. The central findings of this paper are that "[ ] in any
environment in which problems can be hidden by resources, and the learning
is contingent upon mistakes the globally optimum policy will be
operationally suboptimal."

The implications of this finding are that:
. Suboptimal inventory strategies are justified only in
the context of learning and should not be adopted in
. JIT is not intrinsically superior to conventional
inventory management.

There is amazing consistency in these three explanations of learning.

Bill Fulkerson



Learning-org -- An Internet Dialog on Learning Organizations For info: <rkarash@karash.com> -or- <http://world.std.com/~lo/>