RE: Deming and software development

[John Carradice <JCarradice@wcinfo.com>]

Mark,

You have asked some interesting questions.  I have spent a good deal of time
on a similar issue.  Going back to the basics may help you find what you are
looking for.  You mentioned, "The volume of output and quality of output are
both difficult to measure with precision."  When I have been confronted with
similar problems, I have turned to the PDSA or CIC.  It is simple and may
prove to be more effective than concentrating on measuring attributes that
are difficult to measure with precision.  Plan what you want to
accomplish... simply make goals that you plan to achieve.  Do what you
planned and study what you just did.  Compare the results with your plan and
continue to Act accordingly.  I know we probably all know this, but I often
forget/overlook the simplest things and spend to much time on the complex.

In the production of our quality assurance package, we ran into the same
problem.  We couldn't measure the some of the items you listed
("portability,
maintainability, robustness, fitness for use, correctness, efficiency, mean
time 'til failure, timely delivery, and within-budget delivery")  We do plan
on making more versions to get closer to perfection.  A couple of steps we
have already taken:  We took the CIC to the next level by allowing and
encouraging our customers to apply the CIC to our package (get the package
as close to open source as you can).  Plan on making the software
limit-free.  Set up tests and several betas to study what you have
developed.  And the most important issue is to talk to your customers,
whether via a phone call or some sort of survey.  To truly understand the
quality of our software, we must strive to understand the customer better
and better (continuously).

John Carradice
Product Owner
www.wcinfo.com/eqip/
(My opinions are my own. They are not my employer's.)