Discrete Chaos and Learning Orgs LO826

Doug Seeley (100433.133@compuserve.com)
19 Apr 95 04:40:35 EDT

Replying to Tim Sullivan, and others who have asked about what I mean by
"discrete chaos" and what is its relevance to Learning Organizations ??

With the term Discrete chaos, I am referring to the sudden, and highly de-
stabilizing, if not breakdown effects, of random variation in
organizational systems. These phenomena, are I believe, behind much of
the "fire-fighting" modes which swamp managers and organizations from
time to time. Such situations further exacerbate a blaming mentality both
with managers and with operational staff. Many of these situations are
caused in our experience, by the flagrant use of the uniformity assumption
to describe the dynamic behaviour or organizations in systems planning and
especially financial modelling activities.

The problem arises because conventional notions of capacity are too
simple.

Readers of Senge's materials and the earlier Systems Dynamics works are
familiar with the fact that much business culture does not even
acknowledge dynamics as such. What I am saying is that using continuous
systems modelling packages like iThink, Dynamo, Stella and many years ago,
CSMP, one must also take care that in many instances that random variation
in demand and supply, and in reliability of system components, are
modelled and anticipated properly. If this is not done, then even though
the beautiful systemic effects of various negative and positive feedback
loops are modelled, much real world but discrete, dynamics are missed.
Unfortunately, in the financial modelling circles near board-level
decision-making all of this is assumed to be mere operational detail,
neglecting to see that bottom-line profit and quality hinge on such
dynamics.

The basic issue is a capacity question from introductory queuing theory.
It is well known that under circumstances in which there is significant
variation in the times between the arrivals of discrete items, that if the
arrival rate is close to the service rate of the workstation, that huge
congestion can occur. Careful study from our animations of this phenomena
show that the behaviour is highly varied and unstable, virtually chaotic.
[Very often the chaotic effects of a burst of congestion has a recovery
time which is in the same order of magnitude of the burst frequency,
guaranteeing a system breakdown.] Why is this observation important in
organizations??

People normally think of production and service activity in terms of Rates
(items per hour, transactions per day, tonnage per year, etc.). Planning
in corporate culture abounds with this thinking, and continuous systems
modelling tends to reinforce it. People also assume that full resource
utilization and maximum efficiency is the best policy as well (in addition
to human response objections, this is also a blatantly incorrect
assumption on purely technical grounds outside the scope of this
discussion). Hence, planners will often create situations it is assumed
that full resource capacity should be used, even in conditions of rampant
variation such as the very bursty demand for goods and services.
Unwittingly, by this kind of planning they create the very conditions
which cause congestion, often spreading congestion which lead to various
kinds of system breakdown, hence the fire fighting.

We have been exploring software modelling and management consulting
approaches to address such systemic failures in organizations. One
apparently little known fact has emerged. That is, that one cannot speak
of Capacity of a system in which there is significant variation in
dynamics (very few do not), without specifying the degree of system
reliability which one can tolerate at full capacity. Pushing the system
to capacity creates the breakdown effects alluded to above.

We have seen over and over again, the detrimental effects of not
acknowledging these discrete dynamics in organizations, especially when
coupled to performance measurements which do not take the whole system
into account. These effects are not just on human relations, and harmony
in the workplace, there impact upon the bottom line can be terminal.

There are other aspects to discrete chaos which we have also investigated.
In particular, a connectivity avalanche (or collapse) which occurs in the
evolution of connections in random networks.

Hope all of this helps...

Doug Seeley: Compuserve 100433,133 Fax: +41 22 756 3957
"Choice and Chance are One!"

Doug Seeley <100433.133@compuserve.com>