Am I on the Right Track?

[Jonathan Siegel <jmsiegel@yahoo.com>]

I’m not sure that “variation” and “variety”  are
really two different things. I think it goes to the
issue of purpose. Decreasing variation is beneficial
for some purposes, while increasing it is beneficial
to others. 

I have often found Fisher’s Fundamental Theorem of
Evolution helpful in understanding technological and
cultural evolution in organizations, just as it is
helpful in understanding biological evolution. The
theorem states “The rate of increase in fitness of any
organism at any time is equal to its genetic variance
in fitness at that time.” I would paraphrase this: 
“The more variation there is in an organism
(organization’s) adaptation to its existing
environment, the better able it will be to adapt to a
new environment.”  

Reduction in variation is most helpful the environment
is highly stable -- when we have already identified a
problem and found a workable solution. We then want to
make our solution reliable by making it as uniform as
possible. But when we are trying to find a solution to
a problem – and even more so when we are looking for a
new business problem (opportunity) to address – the
more variation there is to our thinking, the more
likely we will be productive. If our resources and way
of thinking are too attuned to making uniform,
reliable solutions  to existing problems, we won’t
have the creative variation needed to adapt to new
ones.  

Researchers of technological evolution have analogized
the problem to climbing a hill in the dark. The
cheapest and surest way to move higher is to grope
immediately around you and find whichever direction is
up, then go there. Unfortunately, this will likely get
you stuck on some low foothill.  If one wants to get
to global rather than local high ground, the best
solution researchers have found is to periodically go
far afield – to some random spot – and see if the
climbing is better there. This random element is a
kind of variation. Since going afield can be very
time-consuming, climbing is more efficient when the
greatest variation occurs at the beginning, then
gradually reduces as the climber gains confidence
(knowledge) that they have found a good solution. Thus
a “stage 0” approach places the greatest variation,
and the greatest cost, up front.  

It’s worth pointing out that the ground around one
shifts while one is climbing, not necessarily
smoothly, but in jerks. Earthquakes are periodically
happening underfoot in mid climb, shifting the
terrain. One has just gotten comfortable with ones
carburetor design and moved into stable,
variation-reducing improvement mode, when all of a
sudden the fuel injector comes along and changes
everything.  

It’s also worth pointing out that many climbers
working together can coordinate their efforts so it
doesn’t have to be so random or costly. Managed
together, as a system, each can cover terrain spread
out in a way that maximizes the overall effort’s
efficiency. This cooperative effort would be spoiled
by treating the climb as some sort of race with a
prize going to an individual. Behavior which maximizes
a climber’s individual chance of success results in
duplication and interference which greatly reduces the
group’s overall efficiency. 

We can thus have some consolation that there were over
300 automobile companies at the beginning of the auto
age, doing everything from steam engines to electric
cars, and that similarly wide variation occurred at
the beginning of the home computer era and the
internet era. A wide variety of designs is needed to
test out which one succeeds. Ultimately a much smaller
number wins out, benefits from uniformity and
standardization are realized, and the industry is
stabilized. It happens as every new industry is born,
evolves, and matures. 

But if one took a look at the process of new industry
birth, development, and maturation as a whole, one
could see that it could be made much more efficient by
recognizing and managing it as a cooperative endeavor,
with all participants sharing in the risks and
rewards. Great inefficiency is created by giving the
one outfit that happens by chance to produce the
eventual stable design the reward that was actually
earned by the entire endeavor together.  There is then
no incentive to cooperate or to manage the process as
a whole. 

An analogy I find helpful here is the wave/particle
analogy. The stabilization of a new industry and its
adoption of standards etc. is somewhat analogous to a
particle emerging from a wave after passing through a
process. The whole effort, spread out like a wave, is
always needed to produce the eventual successful
“particle.”  We can think of the “winning” climber as
the “particle” emerging from the “wave” of group
climbing activity. In complete darkness, it is mostly
luck which one will happen to get to the highest
ground. So too the “winning” company. Indeed,
cooperation and competition create patterns of
interference which can be directly correlated to
properties of a wave. Likewise, the further the wave
spreads, the better the single eventual particle is.
Each element of the wave thus contributes to the
eventual particle, whether we perceive this or not. I
am therefore increasingly thinking of this way of
explaining evolution as being something real rather
than a mere analogy.  

One can certainly understand why people would think
that those who produced the “winning” design must have
characteristics – intelligence, charisma, who knows –
that made their winning inevitable. We certainly seem
to have no difficulty identifying these
characteristics after the fact. But strangely enough,
we can rarely predict winners in new research and
technological endeavors. Who would have thought that
such an obvious genius as Edison would lose out to a
Westinghouse? 

Deming’s key epistemological insight is recognizing
that when a theory can only help us explain something
after the fact, but cannot help us predict it
beforehand, that theory is simply bogus. We should not
believe it. The better explanation is that the
phenomenon is due to chance. 

Applying the result to technological evolution
0requires us to debunk and discard the theory of
individual genius, of picking winners, that has so
mesmerized our society, and so taken over our
institutions of higher learning and business. We can
predict who will succeed in very controlled
environments – who will get good grades by the rules
of academia or score points on the athletic field –
but we cannot predict who will engage in fundamental,
society-changing innovation. 

The better explanation, difficult as it may be to
swallow, is that innovation, like groping in the dark,
occurs mostly through luck. And luck is just another
word for variation. The secret to innovation,
therefore, is understanding, channeling, and managing
variation, making it work for us rather than against
us. A key to that secret is understanding that
innovation is a group effort. Only through a diverse
group, through parceling out variation and managing it
as a cooperative system, can one manage variation in a
way that makes successful innovation more likely and
the whole process more efficient.  

One final point. An organization, like an organism, 
needs to be BOTH be highly reliable -- adapted to its
current environment (minimal variation), AND highly
innovative -- able to adapt to new environments
(needing maximal variation).  A lot of Deming’s
thinking is about how to balance and manage these two
needs of high reliability and high innovation.  

Jonathan Siegel
1859 Shirley Lane Apt A5Ann Arbor, MI, 48105
(734) 994-8089
jmsiegel@yahoo.com 

Through 8/2003:
1060 Carolan Ave Apt 113
Burlingame, CA 94010
(734) 657-1900 (cell)