I was up on a six-sigma
forum not too long ago. Now I was one of the first 100 people
to go through six-sigma training in the world. A lot has
changed, but not everything. Almost all of what I was taught
can be traced back to Deming or Juran and that is still true
today. As a matter of fact, most can be traced back to Deming
or Juran. Some of what has changed works real well if you’re
a large company, but is not
cost efficient if you are a small company. Sometimes, some
of the people at some of the larger companies don’t
understand this. But it is still a lot about common
manufacturing statistics that have been around for years. Some
people have found some short cuts to do things like calculate
sigma, etc. but it is for the most part the same old
stuff.
Things, like the basics, have not and
should not change. For example, 1.5
sigma shift, (or hopefully lack thereof). I see
people ask why not seven or eight sigma. Well, six sigma gives
you 3.4ppm. (Parts Per Million) This is what Motorola decided
they needed to have a 10 fold increase in profits. Why,
because most of Motorola’s products are complex and have
many steps in the manufacturing process. With many parts as
well. It was based on a typical 1200 part step process. There
is no magic to the 1.5 sigma shift as some may want you to
believe. Juran indicated back in 1980 when referring to a 3
sigma wide bell shape curve on a normal distribution,
that if your specification was at 4 sigma, then there was a
better chance to catch bad parts before shipping them when you
were doing SPC,
(Statistical Process Control). While there is no
reason why you could not use seven or eight sigma, it just
wasn’t practical from an economic point of view for Motorola
to do that. At some point spending money to train and monitor
people and processes just wasn’t going to return on the
investment. You CAN have too much of a good thing. Of course
all of this goes ditto for GE, Honeywell, etc.
A lot of concepts of the original six
sigma have been lost, as a lot of you well know. Lets face it,
if you are just stamping out pieces of metal, how many things
could possibly go wrong? Can enough go wrong to justify the
training of a black
belt? If you have a minority company of 10 people, how
many hours could it take to eliminate a process problem? Who
are the black belts going to train, and how many times? We
need to keep six sigma practical by using common sense.
Back to the forum, one guy showed how
you could estimate PI. You know, 3.14159. Now maybe he wants
his auto’s tires built with an estimate of PI, but I think I
would like mine built using the real number. Just because
something can be dome doesn’t necessarily make it practical.
And in some instances, it is just wrong.
Some of the people on this forum are
the nastiest people I have ever run across. I can’t believe
they have had formal six-sigma training. One concept of six
sigma is that there are no bad ideas. If you disagree with
someone, you don’t attack him or her personally. You attack
their ideas maybe. They attack you personally. If this is what
six sigma has come to then I want no part of it. They may know
how to make the measurements and know the theory, but they are
NOT getting it when it comes to the culture part. And six
sigma, perhaps above all is about culture. A culture of common
sense.
