Six Sigma

[Mr John McConnell <Wysowl@msn.com.au>]

In the very first Motorola six sigma seminar conducted in Australia in 
the late eighties, Dr. Harry Michael (not sure about the name’s 
spelling) explained the basics of the six sigma concept he had helped 
develop at Motorola, which he said came in two parts:
1.	Six sigma was a term they developed because most 
characteristics they measured displayed variation such that the specs 
were about three (or four or thereabouts) sigma from the centre line.  
What they wanted was for every one to halve the existing variation so 
the specs were six sigma from the centre line.  Hence the name.  No 
exactitude relating to the normal distribution was meant or inferred.  
It was simply a convenient way of expressing the idea.  Time and again 
he stressed that the term six sigma meant REDUCE VARIATION, which was 
a job for everyone in the business, in every job they did.  Harry 
produced a slide that showed that as variability in the components was 
reduced, first pass yield increased dramatically.  I still have a copy 
of that slide.  He further explained that best practice in their 
industry (the Japanese) had achieved this low level of variation, and 
that they believed this was necessary to become world class and 
profitable.  Merely meeting specs was simply not good enough.  (Sound 
like WED?)
2.	The second element was the use of folk trained in more 
sophisticated statistical techniques to find and target those 
characteristics that gave greatest leverage.  One example he gave was 
the growing of crystals.  Motorola was laser trimming them to achieve 
the desired specs, at great cost.  A multi-variable response surface 
chart showed that if a couple of other variables could be controlled 
to within a given but extremely tight range, they could allow the 
crystals to vary significantly with no loss of performance and 
therefore eliminate the laser trimming.  This they did and dropped a 
lot of cost out of the business.

He referred to the first approach as a “shotgun” approach…something 
everyone did everywhere.  The second approach was the “sniper” 
approach that targeted specific issues.  He even used a slide showing 
a distribution in the cross hairs of a telescopic rifle sight.  I 
think I still have that slide lurking in my archives somewhere.

Of course, all of this was long before the advent of green, black or 
polka dot belts with a blue bureaucratic ribbon.  Are we in danger of 
losing sight of the original aims of the concept?

I liked most of the original stuff Harry showed us (but not all of 
it).  It was lacking the profound knowledge Deming brought, but it did 
have a strong emphasis on understanding and reducing variation as well 
as understanding and addressing interdependent variables by targeting 
those that gave greatest leverage.  Is that not something Deming would 
have applauded?  Sadly, I can’t say that I’m equally impressed with 
much of that which is nowadays paraded under the banner of six sigma.  
Sometimes one is caused to wonder if despite the fact that the 
original idea was brilliant, that the method and the development of 
the original idea occasionally borders on madness.  When I see much ot 
that which is done under the title of six sigma, I am reminded of the 
nelson Funnel Experiment.  Rule four rules!

John McConnell