[KULeuven]
[PMA]
[PMA-projects]
[PMA-Holonic page]
[PMA-IMS/HMS page]
Conlusions of Watchmakers' parable
This simple story reveals that the ability to develop large complex
systems depends on two elements in fundamental and universal ways.
The first element is the environment. How frequently and how
seriously will the environment disturb the developer? Typical
disturbances in the CIM/E domain are:
- Technological evolution (and revolution).
- Introductions of new products.
- Strong and unpredictable fluctuations of the demand.
The contemporary manufacturing environment increasingly provides
these disturbances.
The second element is the ability to develop stable subsolutions
(subassemblies, subsystems) that survive these disturbances. When the
environment disturbs development activities more frequently, the need
for stable subsolutions increases. For the developers of large com-
plex systems, this means that they need modules of appropriate size
that permit the rapid development and adaptation of the overall
system to the environment. Such systems evolve through the gradual
replacement and addition of modules that survive themselves the
changes in the environment (when seen individually) much longer than
the large system itself (without evolution).
A short analysis shows that developers are hitting against the
proverbial brick wall. Consider a system with n components. Assume
that the probability of surviving the next time slot for these
components is p, where p < 1, and that the survival of the components
is mostly unrelated. In this case, the probability that the system
survives the next time slot is p**n. In other words, the probability
of survival for the overall system decreases exponentially with the
number of components. Very rapidly, the expected lifetime of the
system will drop below the level were it is able to repay its
development and maintenance costs when the number of components
increases.
The parable also reveals that hierarchy comes naturally in large
complex systems because it drastically increases the probability of
their emergence and survival. In other words, large complex systems
invariably have a hierarchical structure. However, this does not mean
that it is necessary (or even feasible) to postulate a hierarchy when
starting to develop a complex artificial system. To the contrary, it
means that a hierarchy will/must emerge that reflects hard con-
straints in the environment; it reflects the presence and absence of
stable and suitable subsolutions. Postulating a hierarchy up front
may cause problems instead of solving problems. Developers must look
for possible natural hierarchies.
Furthermore, the parable reveals the limitations of orthodox top-down
development. When developers follow it in a dogmatic way, the
development of subsystems occurs within a well-defined context
because of the top-down approach. In combination with economic
pressures, subsystem developers will minimise both their effort and
the complexity of their subsolutions. Therefore, they design and op-
timise subsystems for a specific context. This results in subso-
lutions that are unable to survive disturbances from the environment
significantly better than the overall system, which makes the emer-
gence of large complex systems in demanding environments impossible.
Experience confirms this: automated manufacturing systems are either
small (for example islands of automation) or enjoy a stable envi-
ronment (for example mass production systems).
Finally, the parable triggered the events that made Arthur Koestler
proposed,
some 25 years ago, the word HOLON.
[KULeuven]
[PMA]
[PMA-projects]
[PMA-Holonic page]
[PMA-IMS/HMS page]
This file is maintained by
Jo.Wyns@mech.kuleuven.ac.be
Created : Aug, 1994
Last update: Jan 5, 1995
URL : http://www.mech.kuleuven.ac.be/pma/project/goa/hsm-int/parab_cl.html