Can anyone define what Six Sigma really is?

[Jen Kirley]

Good discussion as usual.

My view is consistently that while tools tend to morph over time, the degree of their successful use has always been in variation at any given point in the tools' progress. Does that make sense? A sling shot can be an effective hunting tool or not, depending on my skill, discipline, determination and perhaps some other factors. Same goes with the most advanced hunting tools. The user is charged with doing a good job with it or fouling it up.

My understanding of 6S is that it sought to take out some of that variation by improving the structure, the discipline of using the same sort of tools that already existed. It sought to bring better repeatability, which I can agree seems appropriate to seek. While winning its converts it somehow turned into a sort of religion (no disrespect intended) to hold in magnificent ideal.

All in all the success still relies on the user, his/her environment, resources etc. I've seen few organizations that can pull off a 6S effort satisfactorily, and no one who could take the tool box any old where and make it work, because organizations vary so--even departments within them can vary enormously.

This is why I've maintained a suspicious distance between myself and the discipline, and trained a critical eye on the idea that any person with a Master BB certificate can, in fact, go into ABC Corp and get "it" done.

This is also why my signature line is Stealth Quality. In my view, we have the math down and the little charts nailed; Human Performance Management is the last frontier.

 

[Bev D]

Jim: OK a few clarifying questions:

what do you mean when you say that six sigma is a "misbegotten strategy" and it is "complicated"? What about six sigma - besides the shift - do you take issue with?

What is the source of your comment concerning Motorola "changing the meaning of what a defect is"? I've not heard that one before...I don't need cited references, just curious.

I fully admit that I do not do the Six Sigma as advertised by Mikel Harry. Which seems to be the primary source of what most of the vocal people 'know' about Six Sigma. (Even if not written by Harry, most of the articles I have researched stem pretty directly from him.) But Mikel Harry's Six Sigma isn't everybody's Six Sigma. Despite his attempts he doesn't own exclusive rights to it.

 

[Jim Wynne]

what do you mean when you say that six sigma is a "misbegotten strategy"

I mean that its pedigree is uncertain, and its claims are unproven and often shrouded in mystery cost savings that somehow never find their way to the bottom line. Without the shift, even the name "Six Sigma" makes no sense. Sounds better than "4.5 Sigma" though, I guess.

and it is "complicated"? What about six sigma - besides the shift - do you take issue with?

Yes, "complicated" in the sense that it adds unnecessary complexity, all of those silly belt things, and installs its own unnecessary caste system. For example: DMAIC is PDCA, with extra letters that serve no purpose.

What is the source of your comment concerning Motorola "changing the meaning of what a defect is"? I've not heard that one before...I don't need cited references, just curious.

I'm sure I can come up with a reference, because it wasn't a big secret. In their efforts to spread the tails, Motorola found a lot of unnecessarily narrow tolerance bands, and stuff would get rejected internally that didn't need to be rejected. This isn't a bad thing. My point was that a good portion of Motorola's alleged gains vis a vis SS came about by not doing anything except making tolerances more realistic. Not something you need a Grand Master Poobah for.

I fully admit that I do not do the Six Sigma as advertised by Mikel Harry. Which seems to be the primary source of what most of the vocal people 'know' about Six Sigma. (Even if not written by Harry, most of the articles I have researched stem pretty directly from him.) But Mikel Harry's Six Sigma isn't everybody's Six Sigma. Despite his attempts he doesn't own exclusive rights to it.

Like it or not, Harry might not be the father of SS, but he was in the room when it was born. Simply disowning him doesn't help your case. The fact that you seem to be saying that everyone can have their own Six Sigma just reinforces my point--you can't logically say that your SS is "truth" if there's no binding exemplar, and the goalposts are always moving.

 

[artichoke]

Jim,

A very good post ... or should I say 2 posts.

I might also add that I feel the reason for six sigma's "complexity" has primarily been a means of justifying consultants' extra training costs, in more tools. Time would be far better spent understanding the basics. For example, for what I have seen, six sigma "experts" understanding of control charts is abysmal. If anyone thoroughly understood control charts as described in Wheeler's "Advanced Topics in SPC" there would be no need for the up to 40 tools of six sigma. As Ishikawa said, "the seven tools if used correctly, will solve 90% of problems in the workplace". Any six sigma trainee who claims he has done this and is seeking "higher fruit" is kidding him/herself.

Bev,

It's great that you appreciate the fallacy of the 1.5. It's also great that you see that the 1.5 has been used by Mikel Harry and others as the basis for their marketing of six sigma. It's great to see you are not lured by the carrot of 3.4 as so many others have been.

I would also hope that you do not use any of six sigma's fallacies relating to the 1.5 such as:

• six sigma tables
• sigma levels
• a focus on specifications and defect counts
• dpmo calculations
• z calculations
• probability focus instead of Shewhart Charts
• normal distribution assumptions
• plots of normal distributions over histograms

All this is as much nonsense as the 1.5. (I'm sure there's more but this is just some of the stuff that immediately comes to mind). If you remove all this nonsence, what's left - Deming/TQM + six sigma's elitist belts ?

 

[Bev D]

I mean that its pedigree is uncertain, and its claims are unproven and often shrouded in mystery cost savings that somehow never find their way to the bottom line. Without the shift, even the name "Six Sigma" makes no sense. Sounds better than "4.5 Sigma" though, I guess.

true it is a silly name. but we Americans seem enamored of silly names don't we?

and yes the whole cost savings drives myself and others crazy - it wasn't Galvin's original intent to worry about the cost savings and many modern managers now see that we shouldn't always try to dollar-ize improvements that can't be seen directly in the balance sheet. but Juran's and unenlightened finance people’s influence is still strong out there.

Yes, "complicated" in the sense that it adds unnecessary complexity, all of those silly belt things, and installs its own unnecessary caste system. For example: DMAIC is PDCA, with extra letters that serve no purpose.

yes the belt name thing is silly - but aside from that it really is no more than saying you are certified quality engineer...it merely says you achieved some level of skill. and yes there are those who have the paper who can and those who can't from college degrees to ASQ certs to RAB Lead Auditor, to black belt. but abuse is nothing new and can't be used to convict or even indict a whole school of thought - or else we would never get anywhere since none of us is perfect and no methodology is without hacks and snake oil salesmen.

The caste system was mainly implemented or perpetrated by GE and Allied Signal -something many regret now as it seriously backfired on them.

DMAIC is merely PDCA? hmmm. I suppose we could also say that PDSA is also an unnecessary alphabetization of the scientific process? the DMAIC acronym has utility: it's specifically named for the steps in Problem solving (8D is too - no magic just a moniker to remember the steps...) it's a great little common language term: we can instantly communicate where we are on a project by referring to it's phase. if the Eskimos can have 20 something words for snow - different kinds of snow - why can't we have different acronyms and monikers for different types of study that are more specific to the situation at hand than PDCA?

I'm sure I can come up with a reference, because it wasn't a big secret. In their efforts to spread the tails, Motorola found a lot of unnecessarily narrow tolerance bands, and stuff would get rejected internally that didn't need to be rejected. This isn't a bad thing. My point was that a good portion of Motorola's alleged gains vis a vis SS came about by not doing anything except making tolerances more realistic. Not something you need a Grand Master Poobah for.

true but Six Sigma is more than that as you say.

Like it or not, Harry might not be the father of SS, but he was in the room when it was born.

well he was there when they named it...but not when it began. or shall I use the phrase conceived? many many others were there for the start and Harry wasn't a player then.

Simply disowning him doesn't help your case. The fact that you seem to be saying that everyone can have their own Six Sigma just reinforces my point--you can't logically say that your SS is "truth" if there's no binding exemplar, and the goalposts are always moving.

but aren't you saying then that things can't evolve? that everyone must do Lean exactly the same as everyone else and there is only one way to do ISO or you aren't really doing ISO? where is that thread on copy exactly when I need it? Last I checked no one, not even Harry, owned the immutable rights to an unchanging Six Sigma. The Six Sigma police won't come get me if I make improvements.

and my point is that too many people can't or won't see past the silly stuff that Harry pasted on to the program that really have nothing to do with the program or very little (the whole belt thing). How were we supposed to achieve six sigma - even if it was just 3.4 ppm - if not for using solid improvement methods? not all 'defects' can be eliminated by changing the tolerances. and that's why I say that this is the core of Six Sigma - the other stuff is window dressing. I DO like the use of intense training sessions and requiring that the students successfully complete vetted projects - after thorough peer review to ensure that the student has understood and used the tools appropriately. This project defense is where the term black belt came from by the way - it's akin to black belt rules (no holding of punches) and taking your black belt test...I DO like the fact that it involves teaching many people in the techniques - not just quality or even engineering. It improves the knowledge and skill level of all in an organization.

 

[Bev D]

Bev,

It's great that you appreciate the fallacy of the 1.5. It's also great that you see that the 1.5 has been used by Mikel Harry and others as the basis for their marketing of six sigma. It's great to see you are not lured by the carrot of 3.4 as so many others have been.

I would also hope that you do not use any of six sigma's fallacies relating to the 1.5 such as:

• six sigma tables
• sigma levels
• a focus on specifications and defect counts
• dpmo calculations
• z calculations
• probability focus instead of Shewhart Charts
• normal distribution assumptions
• plots of normal distributions over histograms

All this is as much nonsense as the 1.5. (I'm sure there's more but this is just some of the stuff that immediately comes to mind). If you remove all this nonsence, what's left - Deming/TQM + six sigma's elitist belts ?

no I don't use those things either. However I do disagree somewhat with Ishikawa - maybe - maybe - 90% of the Problems can be solved by the simple tools but many of these problems have low defect rates; there are just a lot of them. my experience is that complex variation based Problems have large defect rates. and simple tools aren't enough. DOEs are necessary to get to root cause...control charts can tell you when a process goes out of control but it can't usually tell what the root cause is - even of assignable causes and certainly not common causes (using Shewhart's operational definitions)
and I've never seen a fishbone diagram truly benefit anyone except maybe the makers of large printer paper...

now please note that I am NOT saying that simple methods dont' work - they do and I use them. they are necessary but not sufficient.

 

[Madfox]

Six Sigma is tactical targeting...

The Madfox

 

[artichoke]

However I do disagree somewhat with Ishikawa - maybe - maybe - 90% of the Problems can be solved by the simple tools.

There are some companies in manufacturing that are operating at the cutting edge of efficiency. For these, the basics may not be enough.

In the USA, manufacturing has about 15% of the nation's employment and produces about 19% of the nation's income. The majority of employees, producing the majority of the nation's income, work in areas such as finance, insurance, health, tourism and government.

For the majority of people, the basic 7 tools are more than enough. Most people, even quality managers, have great difficulty even fully understanding control charts. Much of what is currently being taught to the majority, is quite inappropriate. For example, the Motorola Uni site suggests that 3 of the "top six" tools that "every black belt loves" are FMEA, T-Test, and DOE. These are quite inappropriate for most industries outside of manufacturing, in fact even within manufacturing they can be like trying to crack a walnut with a sledgehammer. Many will also suggest that latter two are inappropriate in any analytical study.

The tools used in six sigma projects giving greatest benefits are:
Brainstorming 41.5%
Pareto charts 38.5%
Cause and effect 34.1%
(Quality Digest survey of the top 3 ... sum is not 100%):

In summary, I think Ishikawa is spot on !!

 

[Michaelkoh]

Six sigma is a displined and data driven approach for problem solving in any process using DMAIC.

Tools such as cause and effect diagram, operational definitions, process capability (measure), source of variation studies ( analyze), DOE (full/screening factorial-improve), quality control plan and process flow system (control) are great. Pareto analysis (measure/analyze).

And these are used used with statistical software.

It is confusing at times because users complicated a simple problem with it's tools to fulfil six sigma project quota.

DOE are used when process is stable but need to identify the main influencing impacts and D/PFMEA are used to assign a severity number so that the "biggest" impact issues are being addressed first. These are great in identifying critical to process/quality parameters.

 

[artichoke]

And these are used used with statistical software.

My view is that there is a deal of statistical software that can mislead users. Rather than guiding users, a vast array of options are presented that make it easy for neophytes to be led astray. Help files may give some guidance but users rarely consult these. Users find it easy just to click all the check boxes for the many control chart detection tests, click the check box for transforms, or for superimposing normal distributions on histograms, or curve fitting. All these activities detract from good practice.

It is confusing at times because users complicated a simple problem with it's tools to fulfil six sigma project quota.

For the majority of users in the majority of companies, greatest benefits can be had by keeping it simple. I can just imagine the confusion in service industries trying to use for example, DOE. Most users will find control charts challenging and very few will understand the application of attribute charts. For the 85% of employees who are in non-manufacturing and even for most in manufacturing, the basic tools give most benefits.

Quotas can be very destructive. I had an acquaintance in GE who was told she would never be promoted if she didn't participate in a six sigma project. She objected to the principles of six sigma and the abilities of those leading projects, and she refused.