Introduction

In this episode, Bill and Andrew discuss the shades of variation: meeting requirements, accuracy, precision, and precision around variety. Is reducing variation to zero a good thing? Plus, Bill and Andrew share stories that offer practical ways to think about these concepts.

TRANSCRIPT

0:00:02.4 Andrew Stotz: My name is Andrew Stotz, and I'll be your host as we continue our journey into the teachings of Dr. W Edwards Deming. Today, I'm continuing my discussion with Bill Bellows, who has spent 30 years helping people apply Dr. Deming's ideas to become aware of how their thinking is holding them back from their biggest opportunities. The topic for the day is The Paradigms of Variation. Bill, take it away.

 

0:00:28.1 Bill Bellows: Ooh.

 

0:00:28.1 AS: Exciting, exciting.

 

0:00:33.1 BB: Alright. So let me start off by saying this is episode number eight, and I wanna just make a couple comments about episode number seven, where we talked about "all straw" and "last straw" organizations also otherwise known as "me" or "we" organizations, or red pen or blue pen companies. And I just wanna burst a bubble and say neither one of them, neither organization exists, whether it's all or last or me or we. I view it as a... It's really a matter of which direction your organization is moving, it's a really simple model that I've seen get people to begin to appreciate what Deming's talking about, because I think that contrast is very much like a Deming organization versus a non-Deming organization. But instead of black-and-white thinking, there's really a continuum, and so I think... I just want to say at the beginning, it's really a question of which direction is your organization moving? Another thing I wanna throw out is... I don't think people know, I think absent an understanding of the System of Profound Knowledge, if you're in a last straw organization or a me organization, or a red pen company, I don't think you know that. I think if you become aware of Deming's work, you become aware of what could be. And I liken it to Dr. Deming saying, "How could they learn? How could they learn? The answer is frightening, how could they know?" So I think absent an understanding of The New Economics - Deming's work, I think it's hard to appreciate what you're missing.

 

0:02:11.4 BB: That you're being blamed for the grade, you're being blamed for the red beads. You're being blamed for the weather, if you're the weatherman. And the other thing that comes in mind there with that, "how could they know" is... There's a great video with Peter Senge, which he did a case with Dr. Deming, and there's a blog I wrote about it on the Deming Institute website if you just search for Peter Senge and my name. And you can find the blog as well as the link to the video. And in there Senge is talking about the present state of education systems and very much in this contrast of industrial and post-industrial, and he says, very much what it comes down to is, he says it's the water. He says, "We don't know what fish talk about, but you can be damn sure it's not the water." And likewise, I think people in a red pen company are not getting together. You and I talking about, "Andrew, this system sucks. I'm being blamed for the red beads," and I don't think we're the wiser. Now, if you turn me on to The New Economics. And we started listening to DemingNEXT and we became aware. But absent that, I think we're both frustrated, but we wouldn't know better. Alright, it's on the topic of variation.

 

0:03:30.8 AS: It's...

 

0:03:31.5 BB: Go ahead Andrew, you wanna say something?

 

0:03:32.4 AS: I was just gonna say that... That's where I think Dr. Deming's making the point of the difference between training and education. Education is the idea of bringing outside ideas into your mind, into your business, as opposed to training, which is trying to upgrade skills. And I had a little story of that when I was a head of research at an investment bank in Thailand. The whole job of a head of research is managing all these analysts who are writing research reports on company A, buy company A, sell company B for our institutional clients. And the job of a head of research is to try to manage that schedule. And you know that analysts are always gonna be interrupted and clients are gonna call, the market's gonna do this. So they're very rarely on time when they say that they're gonna finish something. So you're constantly scrambling for the morning meeting, because on the morning meeting you gotta have a story.

 

0:04:22.0 AS: And so that was just the job of a head of research. So I did that really well, managing them and, kind of, all that. And then I went to the number one investment bank, the number one broker in Thailand as the head of research. And I asked them, "So how often do you guys miss?" And they said, "Never." I said, "That's impossible." Because I've spent my whole career managing the flow of analysts. They said, "No, we never miss." When an analyst is gonna be on, they're always on. "And how do you do that?" "Well, we do a three-week-ahead schedule, everybody knows that you are held accountable for being that person on that day. And if you find out that you can't do it, you're gonna talk to someone else and rejigger it and say, hey, could you do Friday? And I'll do Monday the next week?" But they never miss. And I just thought, like the water, I never even knew I could go to a different level.

 

0:05:15.0 BB: Yeah.

 

0:05:16.8 AS: And then I went to a different level.

 

0:05:19.8 BB: Yeah, it's...it's the ability to step back. Alright, so on the topic of the paradigms of variation, I wanna throw out four words. Variety, variation, accuracy, and precision. A variety is, there's red beads and white beads, that's variety. There could be, eight different colors, that's variety, sizes of pants 32 waist, 32 length, that, to me that's variety. As opposed to variation is that a 32-inch waist or a gallon of gasoline, every time you go to get the gallon, you get a gallon of gasoline, it might not be exactly a gallon, that's variation. The reason I throw those out to begin with is that Dr. Deming is known in some circles back in the '80s, he was interviewed by somebody at the, I think at the BBC in England and an interview ends with him, with the interviewer saying, "Dr. Deming, if you could condense your philosophy down to a few words, what would it be?" And I thought, he's gonna say... He is just gonna reject that, that "I can't be condensed." No instead of that, he says, "Reduce variation." And I thought, "Oh, no... "

 

0:06:50.4 BB: So, and there are people alive and well today in the Deming community, who will quote that to me? "You know, Bill, Dr. Deming said, we gotta shrink variation to zero." And I said, "So, is he saying we all ought to be the same size? We ought to be the same skin color? Is he saying that he doesn't like diversity? What does that mean? And same religion?" I mean, you could look at religions as variety, and then you could say within each religion there's variation. So part of what I wanna get at today is what I think is confusion as to what he meant by shrinking variation to zero. So there's variety, variation. Accuracy is that when I get a gallon of gas, is it a gallon, or is it a couple ounces high, a couple ounces low? You go to the gas station, you'll see a sticker on the pump that says that it was calibrated to some standard, when you go to buy a pound of meat, are you getting a pound? Are you getting 15 ounces? And so the National Bureau of Standards is looking at accuracy, are all these things... Is every customer in the United States getting a gallon's worth of milk?

 

0:08:15.3 BB: Now, so that's accuracy. Precision is the idea that you get the same value each time, so I could go to the scale and it measures exactly a pound, exactly a pound, exactly a pound. But is that pound the same pound as the National Bureau of Standards pound? So I could be.

0:08:37.3 BB: Sorry about that. I could get the same value each time, and that's precision, but that's not to be confused with accuracy, so I just wanna throw those terms out. Relative to shrinking variation to zero, shrinking variation to zero which I, for the record, do not believe in. Dr. Deming would say anyone could accomplish anything if you don't count the cost. I think if you start to look at what is the benefit of having less variation versus the cost of that, then we can get to some point that makes sense economically as in The New Economics. But this idea of driving defects to zero, driving variation to zero without looking at cost.

 

0:09:24.1 BB: And you can look in The New Economics, we'll come back to this in a future episode. He definitely had in mind that you have to consider the cost, in fact, Dr. Deming would say, anyone could accomplish anything if you don't count the cost. But there's a... What I wanted to reference is a book by Peter Block called 'The Answer to How Is Yes' and what Block talks about is... Could be like, how...we get focused on, we're gonna go off and reduce variation, we're gonna go off and drive variation to zero or non-value added to zero. What Block talks about that I really appreciate, that I think Dr. Deming appreciate is, why? Why did... Let's step back a minute, and so part of what I wanna get at tonight in this paradigms of variation is the 'Why' piece. Okay. So the first example I wanna look at a variation is throwing darts okay? And hopefully that makes sense, you're throwing darts in a dart board and imagine meeting requirements is being on the dart board, so imagine it could be a foot in diameter.

 

0:10:29.4 BB: And in terms of meeting requirements, you wanna be on the dart board. So I throw it three times, and if you get three that are really close together, they may not be on the bullseye, and that says, I'm very precise, but if the three are not on the bullseye, then that's not very accurate. So again, throwing three and getting really, really consistent is one thing, but then how do I move that to the bullseye? So that's an idea that I could first focus on precision, and then often I find that if I could just slightly adjust my release or my arm, then maybe I could then move it over, so I wanna look at that.

 

0:11:14.7 AS: And moving over is accuracy or?

 

0:11:17.5 BB: Moving it over is accuracy.

 

0:11:19.2 AS: Okay.

 

0:11:19.5 BB: I mean, so the first thing could be, I'm just looking for three...

 

0:11:22.5 AS: Get on the board.

 

0:11:23.6 BB: I wanna be consistent.

 

0:11:25.9 AS: Yep.

 

0:11:26.6 BB: And then make the adjustment, 'cause I find often it's easier to make the adjustment, I think it's a lot of work to get consistency. So I just want to separate those out as two different strategies.

 

0:11:39.2 AS: Yeah, just go to the bar and start throwing darts and you'll see it's a lot of work. Yep that helps, that helps, that helps us to understand it.

 

0:11:45.9 BB: Alright, so next. Next I wanna talk about what I refer to as the Two Distributions Exercise, and so here's the context. Imagine that you are in the procurement organization, and your job is to make a decision as to who to buy a given product from. So your company goes out and gets quotes from four different suppliers, and they provide you with the information. And for simplicity, let's say what you're buying are these metal tubes and... Short metal tubes perhaps used in plumbing, they're a given length, a given diameter. And imagine these four suppliers come back to you. And again, you're the procurement person, "Who are we gonna buy from?" They come back and they say, they quote you the price, and they quote you exactly the same price. All four of them quote you exactly a dollar each, $10 each. It's like, "Holy cow, they're the same price."

 

0:12:46.2 BB: Imagine also, they quote the same delivery schedule. So you've got a plumbing supply, you need lots of these, they all tell you they're gonna give you the volume that you need. So I think, "Gosh, volume-wise that's the same, cost-wise, it's the same." Now imagine what they tell you is relative to meeting the diameter, let's say it's the outer diameter is really critical to how these things fit together. And they quote you and say, "All the outer diameters will meet requirements." They're gonna take care of the scrap and they're gonna get rid of the red beads. All the tubes they will send will meet requirements, guaranteed. And you're thinking, "I want that same schedule, same costs, same quality," now what? Well, now imagine they send you the distributions from the control charts and they tell you that these distributions, you're thinking, "Holy cow, these suppliers are using Cisco process control." And they provide you with the histograms, and they say, "These distributions will never change, shape or location." Holy cow.

 

0:13:49.6 BB: And then added onto that is that you're gonna use them as is. So you're not gonna take them and modify them, you're just gonna bring them into the inventory and send them off to the plumbers to use. So you're saying, "Okay, the process is in control, the level amount of variation, location is predictable, stable, forever. How could I go wrong?" And then the last thing they tell you is, procurement that, "Here's the lower requirement, here's the upper requirement, and here's the ideal value." And so then you end up with two distributions. If I was confusing, I meant to say two, not four [chuckle]

 

0:14:24.1 BB: Alright, so imagine you've got two suppliers and the one distribution goes from the lower spec to the upper spec. And let's say it's a normal Gaussian distribution and it starts at the low end, goes up, high in the center, then off to the other, and that's supplier A and then imagine the other supplier uses 10% of the variation, but is towards the upper spec so it's far more uniform, but it's off of the ideal value. And so I've been using those two distributions with people as an ideal scenario saying, "You're never gonna have all that information, let alone that's all the same." And very deliberately, what I want people to do is say, if it's the same price, same schedule, zero defects, guaranteed, distributions never change and you're looking at the lower spec, the upper spec, and you're saying, "Okay, so one distribution, it has more variation, but the average is right in the middle, which is the ideal value. And the other one is shifted towards the high end of the tolerance, but incredibly uniform," who do you choose?

 

0:15:38.3 AS: So it's a tall curve?

 

0:15:39.4 BB: It's a very tall curve, let's say it uses 10% of the variation, 10% of the tolerance and so I've been using that going on 30 years, and I'll have 30 people in the room and I'll ask them to write down on a three by five card, "Who would you buy from?" And I'll say, "Here are the choices you can buy from the, the one that's the widest, we'll call that supplier A and supplier B is the narrow one to the right, or You could say it doesn't matter." And what I find is incredibly consistent inside and outside of Rocketdyne and literally around the world is the majority of people will take the narrow distribution, to the right will call that supplier B, what I ask them, "Why do you like supplier B?" To a person they will say, "It's more consistent, there's less variation." And I say, "Less variation from what?" "Well, less variation from each other." Well Andrew, that's precision.

 

0:16:40.9 BB: And then I ask the others, and my find is three quarters of the room will take that distribution, the one which is precise. And for the ones who are focusing on the wider distribution, where the average is on target, I say, "Why do you like that one?" And they say, "Because it has less variation from the ideal value." Alright? And so I wanna throw that out is part of the confusion I find inside and outside of the Deming community, in the world of Six Sigma quality distribution B, using a smaller percent of the tolerance, is, has the higher process capability index. 'Cause what that index is doing is comparing the amount of variation, the width of the variation to the overall tolerance. And the idea that you're using a smaller portion is valued. And I said, "Okay, well that's not quite the same as what Dr. Taguchi is talking about. What Dr. Taguchi is talking about," and this one we'll get into in a later episode, "is the closer you are to the ideal value, what you're doing is affecting how this is used in a greater system, so if I'm at home cutting a piece of wood to a given length and I want it to be closer and closer to the ideal value, then what I'm gaining is making it easier to put that piece of wood, or whatever I'm making, together.

 

0:18:00.5 BB: And I find that people who preferred distribution B are really confused 'cause in a big way what they're saying is, "I don't care about where I am within, all I care about is using a small portion of the tolerance." And then when I press on that more and more, they say, "Well, I want fewer and fewer defects." I said, "Well, zero defects is guaranteed, so if you really believe in zero defects as the goal, then you should have said it doesn't matter." And so the reason I wanna talk about the paradigms of variation is that one: variation is one of the elements of the System of Profound Knowledge and it's not just the variation in the number of red beads, right?

 

0:18:58.0 BB: And not to dismiss that the variation of the red beads is caused by the system. But what I've tried to bring to these episodes interviews with you is what I learned from Dr. Taguchi is the variation in the white beads and what is the impact of the variation on the white beads. And if we ignore that, then what we're saying is, "As long as you meet print, that's all that matters at the end of the day." And I'd say if that's where you're going then, then you could do the same thing with Lean or Six Sigma operational excellence. What differentiates Dr. Deming's work, I believe in terms of his appreciation of variation as an element of Profound Knowledge, is what he learned from Dr. Taguchi. That the closer we are to the ideal value, that affects how the system, which is another element of Profound Knowledge, comes together.

 

0:19:53.8 BB: All right, so going back to those two examples, what I started to do, one is I was detecting that less variation, less, I was detecting within Rocketdyne and elsewhere that there was a far greater regard for less variation, less variation from each other than being on target. And I was just wanting to one; find out why does it matter if all you have to do is meet spec? Why does it matter? So relative to the paradigms of variation, and this was back into the mid '90s when I was working with some people in manufacturing and was greatly confused over this, and the confusion was, "Is it enough to meet print, Bill? You're not sure? And then we've got these capability indices. We want to use a small portion of the tolerance and then we've got this, "Bill you're telling we wanna be on target, help me understand that."

 

0:20:49.7 BB: Was what these guys were asking for. And the paradigms of variation that I come up with. And I described it, I said, "Well, let's look at it this way." I said, "There's this thing called... Let's call it paradigm A, and Paradigm A is meet print." All that matters at the end of the day, we wanna meet spec. So.

 

0:21:06.4 AS: When you say meet print, print is a kind of a word that maybe not everybody understands what that means.

 

0:21:12.7 BB: Thank you.

 

0:21:12.9 AS: What, that means spec?

 

0:21:13.6 BB: Meet the requirements.

 

0:21:14.6 AS: Meet the requirements.

 

0:21:15.6 BB: Meet the requirements. And so we want the meeting to start anywhere between here and here. And as long as we're in between... So "meeting requirements" such that everything is good, is paradigm A. And so if you went back to those... Looking at those two distributions, if you said it didn't matter which one to take, that would be the paradigm A answer. And that's rarely the case. And so what I was poking at with people is, "You tell me you're striving for zero defects, and then when I give you that information that there's zero defects, why does that not trigger you to say it doesn't matter?" Because there's something else going on. So then the idea that we want incredible uniformity, precision, that's what I refer to as paradigm B.

 

0:22:07.3 BB: And as I mentioned earlier, that is the dominant choice. We want narrow distributions. We want what people refer to as "piece to piece consistency" to be differentiated by the second most popular answer is being on the ideal value what Dr. Taguchi would call the target, which is what I refer to as paradigm C. So in explaining these three paradigms to these manufacturing folks, I said each of them has a goal. So the goal of paradigm A is to meet requirements, but they not only have a goal, they also have an approach. And their approach typically tends to be, "If you're slightly out measure again, if you're slightly in you're good. Can we change the requirements?" And so I thought as... The paradigm A solutions are all about playing with those lines, moving them in, moving them out.

 

0:23:01.1 BB: Paradigm B, which has a lot to do with, I find within Six Sigma quality, is we wanna have a given fraction of a percent of the tolerance. And these indices, the Cpk Cpk, Cp Cpk, and others, there'll be goals of, "It needs to be 1.33 or 2.0, or 1.67, and we wanna strive for Six Sigma quality." Well, the question I ask those people is, "How much money are we gonna spend to achieve Six Sigma quality? And is there a corresponding benefit?" And I don't get an answer. But so the paradigm B approach would be to take the distribution, and try to make it narrower, but narrow to the point that we're only using, 10% of the tolerance. And again, what bothers me about that is that it's not addressing what Taguchi's talking about, which is what we're doing at home.

 

0:24:04.8 BB: Whether it's baking something, we want the temperature to be close to 350 or, whatever it is we're doing. We're, looking for accuracy in how we're pulling something together, is we're looking for an ideal value. And there, what we're trying to do is, as I mentioned earlier, we're striving for, "Can we get precision and then can we make the adjustment to achieve accuracy?" And instead of just saying, "We wanna achieve some given value." To me, what I tell clients I work with and students in my classes is, "What is it gonna cost to achieve precision, to then focus on accuracy? How much money are we gonna spend on that? And what is the benefit?" And the benefit will be improvements downstream, which is looking at things as a system. And what we'll talk about in a future session, looking more at this is examples of things I've been involved with, that address this idea of not reducing variation to zero, but to me it's about managing variation and having the appropriate amount of variation, knowing that it could never be zero.

 

0:25:18.1 BB: But, does it...am I in a situation where meeting requirements is all I need to be. In the world of baseball there's a strike zone. You've got a batter coming up who can't hit the ball no matter what, and you say, "Well, it doesn't matter where it is. Just get it into the strike zone." The next batter comes up. And that batter is very determined to make... And you're trying to get the ball around the bat. Now it depends on where you are within the strike zone.

 

0:25:46.6 BB: Alright. So the other paradigm I wanna get into, and then we'll call it over, is, paradigm D. So there's A, is meet requirements, that's all that matters. B is, I'm looking for precision. C is, I'm looking for precision followed by accuracy. Paradigm D when I explained this to Dr. Taguchi in the late 1990s, and he said, I need to differentiate having one ideal value so I can be working in a place where all the tubes we make are one inch in outer diameter. And, so there's one ideal value, well, maybe what the company is doing is getting into variety and having different outer diameters. One inch, half inch, three-quarters of an inch. And in each case they're looking for accuracy, but accuracy around different values. And that's what Dr. Taguchi would refer to as... Well, he and I agreed to call it paradigm D, which is precision around an ideal value. But depending on your product line, you may have ideal values for different customers. And that's called variety. And so paradigm D is about precision coupled around varieties. So I just wanted to throw that out as well in our session.

 

0:27:16.7 AS: And the risk that you're highlighting is that somebody who's skilled in Six Sigma or some other tools will be patting themselves on the back, that they've got a very narrow distribution in that... And it's inside of spec and therefore they've done their job.

 

0:27:39.4 BB: Yes. Well...

 

0:27:40.1 AS: And what you're highlighting is that there is, there is an additional cost to the business or additional benefit if that narrow distribution could be moved to the target value?

 

0:27:58.2 BB: Well, here's what I've seen. I've seen organizations go from a really wide distribution where, in the assembly process, they need all those different sizes to put the puzzle together. And then somebody comes in and shrinks the variation to a fraction of that, not taking into account how they're used, and instead of going around and having all the different sizes to put the puzzle together, they can no longer do that. So what I'd say, I've seen plenty of examples where a given amount of variation that people are used to, that they're accommodating could be quite well until somebody comes along and gets rid of those other options.

 

0:28:48.2 BB: So I've seen variation reduction gone sour, a few times leading to some near catastrophic failures of a rocket engine because we're just looking at something in isolation. And, so I went to a very senior executive in that timeframe and I said... 'cause there's this big push in the company and we gotta reduce variation, "We gotta reduce variation." And I went to him and I said, "If we have a choice between shrinking the variation and doing nothing, I'd say do nothing." And he is like, "Well, what do you mean?" And I went through and explained this scenario with him and he said, "Oh, I've never seen anything like that." And I thought to myself, "You must have worked for companies that make the tubes, but don't use the tubes."

 

[laughter]

 

0:29:33.4 BB: I said. And so, this is why when I hear people talk about reducing variability, reducing cost, trying to make improvements, and again, we'll look at this in a whole nother episode, is my concern is are they thinking about that part in isolation? Are they thinking about how that fits into a greater system? So whether it's reducing the variation in the outer diameter, whether it's reducing the cost, if they're focusing on that as a KPI, and not looking at how that KPI fits into a greater system, I'd say I'd be nervous about that.

 

0:30:17.4 AS: One of the interesting examples I remember from when I was young and in maybe business school or whatever, was when Toyota came out with Lexus and they talked about how they spent a huge amount of time reducing variation in every part so that you had a much smoother and more quiet ride, and the reliability was better and better. And they talked about the pursuit of perfection was the tagline that they did. But it made sense to me that, many people would be... Many companies are satisfied with a certain amount of variation.

 

0:30:54.8 AS: When if they could get it more narrow around the desired outcome, then the knock on effects, particularly for a new company, maybe for an old company, and the knock on effects basically lead people to go, "Go back we want more variation," because you're screwing up everything downstream. But if you're building an operation where you can get more and more narrow distribution around the target output, the target desired output, then you're bringing benefit all the way down the line for the business. What have I got right and what have I got wrong out of that?

 

0:31:33.2 BB: Well, that's fantastic. And a couple things come to mind. I really appreciate that question. Andrew, if you were to do a Google search for Dr. Taguchi and Toyota, because this idea of being on target associated with what he referred to as the quality loss function, which again, will be a focus of another episode, I'd rather one, look at it as an integration loss function, just to reinforce the idea that being close to the ideal value is about improving integration. And that's it.

 

0:32:12.7 AS: When you say integration, what do you mean?

 

0:32:15.5 BB: Who's gonna use that tube? What are they gonna do with it?

 

0:32:18.1 AS: Okay. So downstream, integrating the process with the downstream.

 

0:32:20.5 BB: And so if I'm not looking at how the doctor fits into the system, how the tube fits into the system. So what I find is in the Taguchi community, people will say, Dr. Taguchi worked with Toyota back in the '50s and '60s. Dr. Taguchi and Deming met for the first time in the mid '50s in India. Dr. Taguchi was honored with the Deming prize in literature in 1960, and they would've met then. Don Wheeler in his books on Statistical Process Control, and inside the cover it will say, "In September 1960, a new definition of quality being on target with minimum variation." So there's all that. So what I've tried numerous times over the last 30 years is searching for documentation of Taguchi's influence on Toyota. I found nothing.

 

0:33:10.7 BB: And, so here I'm flying back from Japan, having gone there while Rocketdyne was owned by Boeing to explain these concepts to people at the Mitsubishi Heavy Industries, which is the largest aerospace company in Japan.

 

0:33:25.1 BB: There was a big partnership going on between Boeing, the division I worked for at Rocketdyne was part of Boeing. And, Boeing's, at that time, largest supplier in the world was MHI. So I was on a study team to go over there to... And I explained these ideas to them. They knew nothing about this. They were focusing on uniform... They were focusing on... Their quality system was precision, not accuracy.

 

0:33:47.6 BB: And I was explaining what we were doing with that. Well, flying home, I was sitting in business class, sitting next to me is a young engineer, flying out of Tokyo. He is Japanese. And now we started talking. Turns out he is a graduate of Cal Poly San Luis Obispo in California working for Toyota at the NUMMI plant. And I explained to him red pen and blue pen companies, he loved it. I explained to him the paradigms of variation. And he says, "Bill," he says, "I'm coming back from working with supplier to get them to focus on the ideal value." He says, "That is the thinking we use."

 

[laughter]

 

0:34:29.2 BB: He says they wanna change the tolerance. And I'm telling him, "No, you've got to hold that target value." So you can search the Internet, you won't find this. And so there's two data points I want to get before we close. So one is that the majority of the flight coming home was me explaining this stuff to him, and then afterwards maintaining a relationship with him and his boss and looking to see if I could learn more.

 

0:34:56.0 BB: But he was... For him to say, "That's exactly what we do." Well then I spent several years poking Dr. Taguchi about his loss function concepts and all, and he said, "No company in the United States uses the loss function." And I said, "Really?" He says, "No." He said, "The leading users in Japan are Toyota and Nippon Denso," now known as Denso, a major supplier to Toyota.

 

0:35:21.1 BB: And I said, "What do they do with it?" He says, well, he says, "Bill, they have a database of loss functions for how different things come together." He says, "They have a database for the impact of variation." And I said, "Really?" I said, "How do they use it?" He said, "They use it to guide their investments." That's what you're talking about, Andrew. But you won't find that on the Internet. I've not found that in any literature.

 

0:35:51.1 BB: So, those are two things that I hold there. I believe Toyota is using this somewhere deep in the organization as evidenced by this young guy. And my interest is to expand that appreciation within our community in The Deming Institute, that it is not about uniformity. It is not about precision. And, that improving precision could make things worse. [chuckle] If you're not focused on accuracy, then the question becomes, "Is every situation worth accuracy?" And the answer is, "No. You've got to look downstream."

 

0:36:29.6 AS: Okay. Now it's time for me to ask the question that was asked of Dr. Deming.

 

0:36:34.8 BB: Okay.

 

0:36:35.9 AS: Explain it in one short sentence. What do you think the key takeaway is from this excellent discussion?

 

0:36:44.8 BB: I think what's really important is the need to manage variation, which is the same thing as Akoff would say, the difference between managing actions and managing interactions. The idea is that how I accomplish my task depends upon how you're using it. And so for me to blindly meet a requirement from you not knowing how you use it, well, whether that's you asking me to clean the table and I don't know anything about the table, you saying, "I need you to meet these requirements."

 

0:37:21.2 BB: You saying, "I need this by tomorrow." And I say, "What do you mean by tomorrow, Andrew? Tomorrow at eight o'clock, tomorrow at nine o'clock?" And so I think what Deming's talking about is if I just blindly take a set of requirements and meet them in a way that I interpret without asking you for clarification, is not teamwork.

 

0:37:41.7 AS: Great.

 

0:37:44.1 BB: So I need to know how you're using this.

 

0:37:47.1 AS: And, that's a great lesson. And I think what it's telling us is the idea of communicating and cooperating and getting to the next level has to do with really understanding what the next process is doing with it, and how what you're delivering could be improved so that the improvement is measured by a benefit to the next and the next and the next profit process. Not as a loss to the next one, which is what you explained about if variation got reduced, all of a sudden people weren't built for handling that.

 

0:38:23.2 BB: Well, and let me throw one other thing out along those lines. And as a colleague of mine in Amsterdam says to people in the Lean community says, "How does Lean...how does implementation of Lean explain why we love Toyota products? How does it explain the reliability of the products? We buy nothing but Toyotas." Now, we've had bad luck with Toyotas, which people I met in business school classes told me, "You never buy anyone's first model even Toyota."

 

0:39:03.8 BB: So we will only buy Toyotas, but we'll never buy the first model year. And I'm buying it because I want it to start every single time. I don't want a car where I've gotta replace the water pump. And so for our listeners, if you wanna have customers revere your products for the reason, I think, many people revere Toyota products, I think what we're talking about tonight is a significant part of what makes those parts come together and those cars last so long.

 

0:39:41.3 AS: Bingo. Bill, on behalf of everyone at The Deming Institute, I want to thank you again for the discussion. And for listeners, remember to go to deming.org to continue your journey. This is your host, Andrew Stotz, and I'm gonna leave you with one of my favorite quotes from Dr. Deming, which is, "people are entitled to joy in work."