fyi ... attached is a review of this book. I wrote it last year for
an internal HP newsletter.
Joe
====================================================================
Kevin Kelly, Out of Control; The Rise of Neo-Biological
Civilization, Reading, MA: Addison-Wesley Publishing Co., 1994
--------------------------------------------------------------
Kevin Kelly is a journalist, someone who intelligently observes
and communicates. He is currently editor of Wired magazine,
which is kind of the trumpet and cymbals of the Silicon Valley
publishing orchestra. His observations have motivated him to
write an equally noisy book, a dense, meandering productions
that calls into question much of what we now cherish as
leadership and management. He says that our technical and
social children have become too big to control. Successful
complex things can not be built; they must be grown.
Kelly's basic thesis is that "more is different," and we will
inevitably fail when we try to solve big problems using the
techniques that worked on small ones. He says that the best way
of dealing with large, complex systems, whether technical,
social, or biological, is to let go of them, to let them grow in
an evolutionary manner rather than indulging in the hubris that
we have the knowledge and foresight to plan and engineer them.
His model is the "swarm" or the "network," and he describes the
rules he sees for adapting to this reality.
Kelly offers a simple/not easy formula for developing complex
systems:
1. Do simple things first
2. Learn to do them flawlessly
3. Add layers of activity over the results of the simple task
4. Don't change the simple things that are already working
5. Make the new layer work as flawlessly as the simple
6. Repeat as needed
Every engineering bone in my body screams in protest. Think of
the rats-nests of inefficiency I'm going to bury in those
layers. Think of the patches, the unused code, the go-to's.
Yuk!
All for the sake of having something that does productive work
soon and that is always returning benefits even while
environments are changing. Hmm... I may hate it, but my
business users are going to love it.
I think I'm smart enough to plan for future changes, along with
my very smart users. I want to give my user managers control
over their processes. But I've never seen a really large
project work right the first time, even my own. Some famous
ones, like the Denver airport, become fiascoes and revenue
faucets for liability lawyers. Others, like the Panama Canal or
the more recent English/French Chunnel burn out generations of
financiers, politicians, and engineers before they learn enough
to get the task done.
Kelly spends most of the book talking about the role of two
kinds of evolution, the Darwinian survival of the fittest, and
the earlier discredited ideas of Jean-Baptiste Lamarck who wrote
about the effects of life on the environment and vice versa,
effects some people now call "learning." Kelly feels that
Darwinian evolution of species through adaptable mutations is OK
for long (reengineering) cycles, but Lamarckian ideas of
evolution need to be applied for what we in the quality world
call "continuous improvement." And Kelly (and Darwin) admit
that the theory of evolution in any form still has its blind
spots, since we still have no good theory to explain the
appearance in biology of sophisticated "subassemblies" such as
eyes.
Kelly's excursion finally arrives at Chapter 24, "The Nine Laws
of God." The first, topic sentence of the chapter summarizes
the book: "Out of nothing, nature makes something." And here
are the nine laws nature follows to do that great trick:
* Distribute being. Use hive, swarm, and network organizations
rather than hierarchies.
* Control from the bottom up. Overall governance must arise from
the most humble interdependent acts done locally in parallel,
not from a central command.
* Cultivate increasing returns. Use skills; strengthen them; use
them again. Use positive feedback. Confidence builds
confidence.
* Grow by chunking. Begin with a simple system that works. Take
time to let each part test itself against others. Create
complexity by assembling it incrementally from simple modules
that can operate independently. (Doesn't this sounds like HP's
division model?)
* Maximize the fringes. Look in the hidden corners, the moments
of chaos, the isolated clusters for the hints of future
innovations.
* Honor your errors. Learn from them. Evolution can be
described as systematic error management.
* Pursue no optima; have multiple goals. Rather than striving to
optimize any one (much less several) functions, complex systems
survive only making lots of functions work just "good enough."
* Seek persistent disequilibrium. Operate at the edge of
stability. Think like a skier. Standing still is just as bad as
a tumbling fall.
* Change changes itself. Complex systems are kept alive by
adapting, by having mechanisms to deal with change and to go one
level higher, to change how changes are made.
...
So where does all this leave us? We're in a big, complex
company. We're awash with tough, rapid changes. Our old
successful processes aren't working. They don't scale up. One
famous definition of insanity is doing the same things over and
over and expecting different results. We need a new way of
looking at things, a new paradigm. Is evolutionary methodology
an answer or a cop-out?
I'm not sure that it's the answer, but it can't hurt. Getting
small, early successes is sure better than having large, long
failures. Someday, we may find a way of doing large, complex
projects right the first time. But until we do, evolution sounds
pretty good to me.
--
Joe Podolsky
podolsky@corp.hp.com