Zero Defects & Taguchi Loss Function - ISO 9000 and the Zero Defects Philosophy

[Marc]

Six year old sez: "I want to be President!"
Goal is set.

Works like hell all his life to achieve goal.

At age 60 has not reached goal. May or may not still be trying to reach goal.

------

Company sez: "I want zero defects!"
Goal is set.

Company (if 'smart') chooses 'path' to goal.
(Roadmap & Methodologies, if you will).

Company never achieves goal (Maybe - We, today, believe zero defects is not a possible reality - but, man wasn't meant to fly, either... Not to mention cloning). May or may not still be trying to reach goal.

------- My Comment To All This -------

I see zero defects as a goal. A philosophy maybe. Concept? Ummmmm....

If you embrace Continuous Improvement, your goal (ultimate) is zero defects. I thank Crosby for the thought - the statement, if you will - like ISO9000 it's common sense, really - but I prefer to set my own roadmap, thank you.

I see no problem in using it as a goal but it's in the actions. You don't have to shout out "Our new Program of the Month Is....". To me it is more of an ultimate goal. You know, like "Some day, before I'm 60 years old, I'm going to get to Tahiti and spend 3 years lying with a wahine." As you live your life (design, production, etc.), you do what you can (various Continuous Improvement methodologies) to someday achieve your goal of Tahiti & a wahine (zero defects). Your every day life is impacted by what you do to reach your goal and vice versa (ie: they have inputs and outputs to each other). Same with a company.

Does this make sense to any of you folks???

 

[Batman]

Yes, I think that is what I was trying to say, but said it so much more elequently. I do deal with black/white folks who say "I either plan for no defects" or "I am going to plan for defects."
Yes, much more elequently.

 

[Don Winton]

Marc,

Well said. I know this thread is getting pretty long, but I saw this and thought I would toss it in:

-------Snip-------
Subject: Re: Crosby vs. Deming
Resent-Date: Sat, 30 Jan 1999 03:07:51 -0500 (EST)
Resent-From: [email protected]
Date: Wed, 27 Jan 1999 22:14:10 -0800
From: Rip Stauffer <[email protected]>
Organization: Naval Leader Training Unit
To: [email protected]

Robert Bacal wrote:

"I suspect that in order to discuss zero defects and it's value or lack thereof, one needs to consider it the way Crosby intended it, and that means understanding his definition of quality."

Thanks again, Robert, for raising the hard questions and making me recheck my references. I don't think Crosby ever minced words. On p. 15 of my copy of "Quality is Free" (New York: Penguin. 1980) he defines Quality:

"...we must define quality as 'conformance to requirements' if we are to manage it...All through this book, whenever you see the word 'quality,' read 'conformance to requirements.'...If a Cadillac conforms to all the requirements of a Cadillac, then it is a quality car. If a Pinto conforms to all the requirements of a Pinto, then it is a quality car." (I know Toyota execs must have been praying that Ford and GM hired Crosby and not Deming if they read this!)

He goes on,"...quality is precisely measurable by the oldest and most respected of measurements--cold hard cash....Quality is measurable by the cost of quality which, as we have said, is the cost of doing things wrong."

I had occasion to re-read QIF a couple of months ago when I was trying to figure out the origins of Six Sigma (please, let's not get started on THAT.) I think his ideas would have been pretty good back in the 40s, maybe, and I wish we had at least been doing what he suggested when I was at Ford in the '70s. It would have made Ford a much better place to work, and probably would have raised the level of quality we were building at that time.

It would not have solved the bigger problem, though--the problem of Japan. Ford management in the '70s did not have a clue as to why Japan was making the dramatic inroads it was making. We could have made Zero Defects all day long, and would have been ahead of everybody that had not come as far as we had, but we would never have become competitive with an industrial culture whose understanding and definition of quality had shifted for good in 1960. Taguchi redefined world-class quality in 1960 as "on target with minimum variation." Crosby's definition, probably valid and valuable in the earlier part of this century, was woefully obsolete and utterly impotent after 1960 (I am fairly certain that I could make a case for 1931, but that would probably be a full-blown paper).

I can't find anywhere in Crosby's book where he goes beyond conformance to specifications in his idea of what quality is. He talks about putting inspectors at the end of the line--they are not effective anywhere else. He says, "If you don't know what the defect level is, how do you know when to get mad?"

He says a lot...some pretty good, some pretty practical. A Crosby seminar was a kind of fun thing, and DID get everyone revved up about quality for a while. But Zero Defects is just no non-conforming product. If you're doing that in a world where meeting specs is enough, that's good--but we haven't lived in that world since 1960 (or 1950, or 1931, if you prefer).

I'm sorry. I tried, but I just could not find a way to reconcile Crosby's approach with this idea. My hat is off to anyone who can somehow make a case that I am failing to understand what Crosby meant by anything in QIF. If you can establish that there is some way to read an understanding of variation, systems, and the concept of the loss function between the lines in QIF, you are a much greater master of tricky semantics and verbal legerdemain than I.

I hope we can soon leave this discussion behind. I think that if we are eventually to go beyond Deming, we should not waste a great deal of time in dismissing an approach that stopped well short of Deming.

Rip Stauffer
Naval Leader Training Unit
[email protected]
-------End Snip-------

Regards,
Don

 

[Marc]

Damn... Another list I should be reading....

Good posting, Don. I guess I hear stuff from all sides and take my own view which is why I said above that "I'll choose my own roadmap, thank you." As to the definition of 'quailty' - well, we could all write papers on that!

Taguchi's loss function is a bit more concrete to me than Crosby's Zero Defects.

 

[Kevin Mader]

Don,

That was a great post. Thanks. Kev

 

[Don Winton]

I'll choose my own roadmap, thank you.

I agree Marc. I have never been one to accept any so-called "best" way at face value. I prefer to design hybrids based on a particular organization's wants and needs.

As to the definition of 'quailty' - well, we could all write papers on that

Wouldn't be a great read!

Taguchi's loss function is a bit more concrete to me than Crosby's Zero Defects

Agreed. Also easier to explain in terms most can relate to.

Regards,
Don

 

[Smooth]

I have heard that Taguchi's loss function has certain defects and that a modified loss function is developed. I have yet to find any literature on this matter. How true is this and any idea where I could get literature on it?

Regards,
Smooth

 

[danqa]

Loss Function

I still see some efforts with the Taguchi Loss Function. Has anybody noticed that it's counter to 6-sigma, and does not really 'speak' the language of management -- "real" $? He speaks of 'Loss To Society". Admirable as that may be, my experience tells me that management is first concerned with offering a good product at a good value at the best price -- and to constantly reduce costs to enhance shareholder's equity.

For example:
In Taguchi's book, Into to Q E, on page 22, he presents a problem:

".... Customer Tolerance is m +/- 200; Cost of exceeding the tolerance is 8000. Factory Cost is 300; process standard deviation is 40.

According to the loss formula, the process tolerance should be set to 39 (later modified by a recursive relation to 45 on page 23) -- resulting in a 26.7% percent defective!"


Let's analyze this situation:

Doesn't Make Sense Item 1:
I take a process with a Cp/Cpk of 200 / 3(40) = 200 / 120 ~1.67. AND the Defect Rate is <100ppm; AND now I reduce the tolerance to the supplier to 45 and incur a loss to the supplier with a defect level of 27%???? With a statement that we need to work with the supplier to reduce his tolerance?

Does anybody see that this is not right? I can't convince a manager (nor should I attempt to) to reduce a tolerance by this formula! My process is Cpk of 1.67! Changing this process would be a low priority!



Doesn't Make Sense Item 2:
TECHNICALLY, it will always be cheaper to fix a problem internally rather than at the customer's site. Thus, the reduction ratio will always be less than one. And if I leave the company, the next person can come in and re-adjust tolerance even lower. There is no stopping criteria, because the formula is based solely on the ratio of expenses -- and presumably, that ratio of fixing the probelm at the customer versus internally will remain the same. Thus, here comes a new person, and reduces even further.




Historical Note:

The loss function was developed in the 50's and it was developed around both a linear/quadratic model. The key item for managers was to establish a breakeven cost point, and adjust the process in accordance with a true breakeven.

I can give you a correct formula that can actually be used to communicate the correct decision based on business decisions and "real" dollars -- which does effectively communicate the language of management.

Thanks,.....


danqa

 

[Sam]

Please post you equation.

 

[cncmarine]

I see zero defects as a goal. A philosophy maybe. Concept? Ummmmm....

A Goal is a Dream with a Dead Line.


All Crosby is doing is preaching what we all ready should know.

The objective is Zero Defects by the use of cross functional teams. Also empower the employees on the manufacturing floor. (Just like 5 S)

I don’t agree with all that’s he says but at least is more beneficial the “shalls” in 4.6 Purchasing