Truth of Six Sigma

[S Santosh Kumar]

"To achieve Six Sigma quality, a process must produce no more than 3.4 defects per million opportunities."

Does it sound reasonable. I think it is just sugarcoating the basic principle "bug fix" for wooing customers.

Kindly comment !!!

[Roger Graff]

Can this really be applied to software development? It sounds to me as if that system was developed for manufacturing and someone is trying to square peg it into a round hole.

How would one define an "opportunity" in terms of code?

[Dmitry Melnik]

How about having a big and complex software system fail 3.4 test cases out of 1M ?

BTW, at a certain phase of the whole process, software development is a kind of manufactoring. You write the code according to specs in the same way as cars are being made according to blueprints. If you look for similarities you'll find them a plenty.
[ July 21, 2004: Message edited by: Dmitry Melnik ]

[S Santosh Kumar]

Six Sigma is for Software also. You can search and find a lot of companies claiming to be Six Sigma company.
In practical situations, is it possible to have 3.4 or less bugs in millions cases ???

I doubt

[Tim Holloway]

Originally posted by S Santosh Kumar:
Six Sigma is for Software also. You can search and find a lot of companies claiming to be Six Sigma company.
In practical situations, is it possible to have 3.4 or less bugs in millions cases ???

I doubt

It's easy. Just count bugs per bit!

[Dmitry Melnik]

In practical situations, is it possible to have 3.4 or less bugs in millions cases ???

May be. But it costs some time and money. Would it be practical to shoot for 3.4 defects per 1M, in a situation where 5 defects per 2K keep the most demanding (current and prospect) customer happy and pleased?

On the other hand, I wonder how large and complex software systems are being tested. Like operating systems, or processor chip layouts (which are I guess big programs written in VERILOG) What are appropriate failure rates in different banches of sofware industry?
[ July 22, 2004: Message edited by: Dmitry Melnik ]

[Mark Herschberg]

Originally posted by S Santosh Kumar:
In practical situations, is it possible to have 3.4 or less bugs in millions cases ???

Know anyone using a pacemaker? Ever fly on an airplane?

--Mark

[Mike Gershman]

Pacemakers and airplanes have had design errors, even fatal ones.
Programs rarely have manufaturing errors.