A system is irreducibly complex if
a) it consists of several parts,
b) all the parts are needed for the system to function
c) no development path to the present form from a simpler form consisting
of functional intermediates exists
Mike then claims that irreducible complexity is strong evidence for
intelligent design.
Glenn presented a counterexample from economics (I formulated a similar
one, but won't bore you with repetition).
My counterexample comes from product development. My previous employer was
a company called Moog, Inc. in East Aurora, NY. One of their divisions,
Hydra-Point, which no longer exists, made NC (Numerical Control) and CNC
(Computer Numerical Control) machine tools. For a period of time I worked
with the group that developed controller software and hardware for their
first CNC machine. Hydra-Point's process for developing and producing
machine tools had grown by trial and error for a number of years. Many of
the mechanical designers weren't mechanical engineers. Some electronic
design was done by people who had little or no formal education in
electronics. Important decisions were made in the most informal of
meetings and often not documented. These people hated paperwork and
avoided it when possible. In spite of this they produced some pretty
decent products. However, implementing changes frequently caused
incredible disruption in company operations. We usually managed to protect
the customers from the fallout, but it wasn't easy. Our system for
producing machine tools had evolved and had developed complex, interacting
processes that weren't documented. Any attempt at change generally
produced several textbook examples of what economists refer to as the law
of unintended consequences. Attempts to plan changes carefully ran into
mazes of interlocking considerations that made it seem impossible to
implement change. That comes pretty close to Mike's definition of
irreducible complexity. Yet it didn't result from design. Rather it
resulted from _failure_ to design.
Admittedly this example doesn't satisfy all of Mike's criteria exactly.
However, I believe it's close enough. The chief discrepancy might be in
c). One might object that since we know that the company existed for 30
years or so and updated its product development process to reflect new
technologies and new products over that period, that obviously the process
had developed through a number of more or less functional intermediates.
However, I will show that this example suffers from the same sort of
difficulty with c) that Mike's do. It's clear in this case that a path does
exist through functional (mostly) intermediates from the company's first
approach to development to its present one. But allow me one change to
make the example a better analogy of the intelligent design problem: Let
us assume that we all we have available to study is the current state of
the company's development system and fragmentary information about
development approaches of the past(actually that's not far from the truth.
Remember these people hated paperwork). If this were the case, it would be
unlikely that we could reconstruct the path of functional intermediate
development approaches used by the company. In particular, at times in the
past where new approaches were tried that had problems, it would be
difficult to reconstruct the history of the product development process.
In those situations, various solutions would have been tried in rapid
succession, with the failures discarded and not likely to be documented.
New approaches would have run into unanticipated difficulties, and trial
and error, intuition and luck would have contributed to the solution that
was actually adopted, which may have looked very different from the
original plan. In retrospect it would be difficult to show that aspects of
the product development process of 1982 developed logically from the
process in use in 1970. That's exactly the difficulty with Mike's
definition of irreducible complexity: showing that c) holds is not
possible.
A second counterexample might be the functioning of a moderately complex
social system. Again we're looking at a system that has grown to its
present state with little or no conscious overall design. By trial and
error the system had found operating points -- or perhaps a trajectory of
operating points -- which work pretty well. But attempts to change such a
system run into hordes of problems because it is difficult to assess in
advance what the effects of seemingly simple changes will be. I claim that
this is an irreducibly complex system that is not designed.
Bill Hamilton
-------------------------------------------------------------------------------
William E. Hamilton, Jr, Ph.D. | Staff Research Engineer
Chassis and Vehicle Systems | General Motors R&D Center | Warren, MI
810 986 1474 (voice) | 810 986 3003 (FAX) | whamilto@mich.com (home email)