KM Metrics LO21905

John Gunkler (
Mon, 14 Jun 1999 09:38:22 -0500

Replying to LO21895 --


I appreciated your reminder that the organizations for which knowledge
management is most necessary are, almost by definition, complex ones. I
also know, from my own experience, that it is never the case (and I
usually avoid the word "never") that only one thing will change in a
complex organization -- therefore, that there will always be several
candidates for credit or blame when the bottom line changes. [By assuming
an experimentalist's stance, it is possible to arrange change efforts in
such a way as to make implausible the arguments that other changes were
primarily responsible for results. Most organizations don't make the
effort to set up "experiments," though. The closest they come is "Let's
try it and see what happens." But this casual approach won't allow
inference from effects to causes.]

However, simply because a system is complex and non-linear doesn't
necessarily mean that one cannot implement a single, simple change and see
a major change in results. So, I must quibble with your statement (below)
although I sympathize with the spirit of it:

>This is attributable to the complexity of human organizations, and the
>enormous variety of OTHER variables that conspire together (with
>knowledge being only one of them) to cause change in a decidedly
>nonlinear fashion. To manipulate only one of them, and to attribute
>whatever change that follows to that singular action, is to engage in a
>form or self deception that I suspect will not last long.

One of the classic descriptions of chaotically complex systems is that
they are extremely sensitive to initial conditions -- or that the flapping
of a butterfly's wings in South America changes the weather in Minneapolis
many weeks later. Doesn't this say that if someone, intentionally,
changes the right initial condition that large effects can occur?

Also, when one looks at system dynamics models of complex systems with an
eye to improving some aspect of system behavior, often there are one or
two "policy" changes that will (without altering anything else -- simply
letting the system do its thing) change the behavior in question.

The spirit of what you wrote, however, is not violated by this. Because,
in complex systems, cause and effect are usually separated in time and
space. Therefore, the change that causes other changes will work itself
through the system over a (at least relatively) long period of time and
through the mechanisms of many other parts of the process. Because of
this, when the "bottom line" (or other system behavior) change occurs, it
will not be obvious (nor easy to prove) that the policy change one made
was the cause. And I'm talking in an idealize (for example, simulated)
system! In real life, other changes will have occurred during the period
when your change was propagating through the system, and all of these will
step forward as candidates for credit -- so long as the bottom line went


"John Gunkler" <>

Learning-org -- Hosted by Rick Karash <> Public Dialog on Learning Organizations -- <>