In this episode, Bill Bellows and Andrew Stotz explore the intersection of variation and quality through awareness of the "Paradigms of Variation." In a progression from acceptability to desirability, Bill created this 4-part model to offer economic insights for differentiating "Zero Defect" quality from "Loss Function" quality," with the aim of avoiding confusion between precision and accuracy when desirability is the choice. Learn how to decide which paradigm your quality management system fits into! TRANSCRIPT 0:00:02.5 Andrew Stotz: My name is Andrew Stotz and I'll be your host as we dive deeper into the teachings of Dr. W. Edwards Deming. Today I'm continuing my discussion with Bill Bellows, who has spent 31 years helping people apply Dr. Deming's ideas to become aware of how their thinking is holding them back from their biggest opportunities. This is episode 7, The Paradigms of Variation. Bill, take it away. 0:00:30.3 Bill Bellows: Thank you, Andrew, and welcome to our listeners, as well as viewers, if you have access to the viewing version. Yeah, so I went back and listened to Episode 6. I'm going out bike riding 2-3 hours a day, so I listened to the podcast, listened to other things, stop and write down. Let me go write that down. And, so, we're going to pick up today on some major themes. And, what I keep coming back to is, is I think the difference between acceptability and desirability is the difference between how most companies operate and how a company inspired by Dr. Deming would operate. 0:01:29.3 BB: And, I just think of, if there was no difference between the two, then... Well, lemme even back up. I mentioned last time we were talking about why my wife and I buy Toyotas. And, yes, we've had one terrible buy, which I continue to talk about. [laughter] And, it's fun because it's just a reminder that even a company like Toyota can deliver a really lousy product, which we were unfortunate to have purchased. And, we're not the only ones that, and they've rebounded and they've apologized, they've had issues. There's no doubt about that. They have issues, but they have notably been inspired by Dr. Deming. 0:02:30.6 BB: The one thing I brought up last time was relative on this thinking of acceptability, desirability, where acceptability is looking at things and saying it's a quality system of good and bad. It's acceptable, which is good and unacceptable is not good. And, that's how most organizations view quality. Again, the focus of this series is Misunderstanding Quality. Our previous series was broadly looking at implications for Dr. Deming's ideas. And, here our focus is quality. And, so what I'm trying to get across here is quality management, traditional quality management. 0:03:17.4 BB: In most organizations, in all organizations I've ever interacted with is acceptability basis, good parts and bad parts. It's a measurement system of it meets requirements, we ship it, if it meets requirements, we buy it. And, I'm not saying there's anything wrong with that, but I don't think a system focused on acceptability can explain... To me, it does not explain the incredible reliability I have personally experienced in Toyota products. 0:03:46.9 BB: Now, I'm working with a graduate student and I wanna pursue that as a research topic in the spring, 'cause for all I know, the reliability of components in all cars has improved. I don't know if it's, I only by Toyota, 'cause so this woman I've met recently and I'm mentoring her and we're working on a research project. And, I thought recently, I'd like... And, I'm not sure how to do this, but I just know, I think I've mentioned I worked at my father's gas station back in the '70s and I remember replacing water pumps and alternators and all this stuff. This was before Japanese cars were everywhere. There were Japanese cars, but not like you see today. 0:04:33.3 BB: And, so I'm just used to all those components being routinely replaced. And, all I know is I don't routinely replace anything but the battery and the tires and change the oil. I think that's about it. Everything else is pretty good. But, I do think the differentiation between Toyota and most other companies is their appreciation of desirability and how to manage desirability. And, that's why I keep coming back to this as a theme for these sessions. And, what I think is a differentiation between a Deming view of quality and all other views of quality. What I tried to say last time is I just give you indications of a focus on acceptability. It's a quality system which looks at things that are good or things that are bad. It's, last time we talked about category thinking. It's black and white thinking. If the parts are good, then the mindset, if they're good, then they fit. 0:05:38.4 BB: Well, with a focus on continuum thinking, then you have the understanding that there's variation in good. And, that leads to variation in fit and variation in performance. And, that's a sense of things are relatively good, not absolutely good, whereas black and white category thinking is acceptability. They're all good. And, if they're all good, then they should all fit. I was, when I was at Rocketdyne, met, and the one thing I wanted to point out is... Again, as I said in the past, so much of what I'm sharing with the audience and people I've met through these podcasts or people I'm mentoring, helping them bring these ideas to their respective organizations or their consultants, whatever it is. 0:06:29.0 BB: And, so I like to provide examples in here for things for them to go off and try. You at the end of each podcast encourage them to reach out to me, a number of them have, and from that I've learned a great deal. And, so one guy was... A guy I was working with at Rocketdyne, he was at a site that did final assembly of rocket engine components. And, so one thing I'd say is the people who... And for those listening, if you wanna find people in your organization that would really value the difference between an acceptability focus and a desirability focus, find the people that do assembly, find the people that put things together. 'Cause the ones that machine the holes, they think all the holes are good. People that make the tubes, all the tubes are good. But, find the people that are trying to put the tubes into the holes. Those are the people I loved working with because they were the ones that felt the difference every day. 0:07:31.1 BB: And, so I was in a workshop for a week or so. And there's two people ahead of me. They came from this final assembly operation of Rocketdyne. And, during a break, I was trying to clarify some of the things I had said and I used, I shared with them an example of how when we focused on not the tubes by themselves or the holes by themselves, that we focused on how well the tubes go into the holes, which has a lot to do with the clearance between them and the idea that nobody owns the clearance. One person owns one part, one owns another. And, what we realized is if we focused on the relationship, what a big difference it made. So I'm explaining it to him and he turns to me and he says, he's like, "Oh, my God," he says, "I've got hundreds of turbine blades and a bunch of turbine wheels and the blades slide into the wheel." And he says, "I can't get the blades onto the wheel." 0:08:31.0 BB: And I said, "But they're all good." He says, "They're all good." But he said, "Well, what you're now explaining to me is why they don't go together. Why I have this headache." So I said, "Well, do you know where the blades come from?" He says, "yeah". And I said, "Do you know where the wheels come from?" He says, "yeah". I said, "Well, why don't you call them up and talk to them?" He says, "There's no reason for a phone call 'cause all they're going to say is, "Why are you calling me? They're all good." So, he just walked away with his head exploding 'cause he's got all these things. 0:09:05.8 BB: And, so I use that for our listeners is if you want to find people that would really resonate with the difference between acceptable and desirable, talk to the people that have to put things together. There you will find... And, so my strategy was, get them smart. Now they have to be patient with the people upstream 'cause the people upstream are not deliberately doing what they're doing to them. So, what you don't want to do is have them get... You want their consciousness to go up but you now wanna use them to talk to the component people. Now you've got a conversation. Otherwise, the component people say, "Why are you talking to me? Everything I do is good." 0:09:51.6 BB: So, I just want to talk at this point, just to reinforce that I think there's something going on with Toyota that is very intentional about managing desirability when it makes sense using acceptability. So, it's a choice. And, so indications of a focus on desirability is when you look at options that are acceptable and you say, "Of all these apples, I want this one. It's the ripest. Of all these donuts, I want this one. It's got the most sprinkles. Of all these parking spots, I want this one. It's a little bit wider than the other. I want this surgeon. I want this professor for this course." 0:10:33.8 BB: All right. So, what we're saying "is of all the choices, I want this one". So, some new ideas I want to get into tonight are the Paradigms of Variation A, B, C, D, and E. Paradigm A we looked at in the past. That's just acceptability. Does it meet requirements or not? The quality focus is achieving zero defects. And tonight I want to get into B and C. The next time we'll look at D and E. In explaining these ideas recently to someone who listened to one of our previous podcasts and were focusing on, he started asking about decision making. And that got me thinking about, of course, I took years ago decision making with Kepner and Tregoe. And there they talk about decisions. We're gonna look, we're gonna go buy a car, go buy a house. We're gonna make a decision. 0:11:29.4 BB: And, once you decide on the decision, you then list the criteria of the decision. And you come up with all the things you want in this decision. And then you look at each of them and you say, "is it a must or a want"? And let's say you're looking at houses. It could be a lot of houses to go look at. What makes this focus on acceptability, it's musts and wants. And must is very much acceptability. So you say: "We're looking for a house that must be one story, it must be in the middle of the block. The house must be in the middle of the block. It must have four bedrooms, must have two bathrooms". So now when you're looking at all these houses, acceptability says "I'm only gonna look at the ones that meet those requirements". And, so now the strategy is to go from hundreds of options down to an order of magnitude less. 0:12:25.1 BB: Now we're going to get it down to maybe 20. Now you look at the wants. So you've got an original list of all the things, the criteria, and you look at each one and say, "is it a must, is it a want"? And what I've just said is the first screening is all the ones that pass the must get into the next category. Well, with the Kepner-Tregoe folks, they talk about must, which is acceptability, and the wants are about desirability. 0:12:51.4 BB: And then here it ties into Dr. Taguchi's mindset, and we'll look at Taguchi in a future session. Taguchi looks at a characteristic of quality, such as the diameter of a hole, the performance of an automobile, miles per gallon. And he says, in terms of desirability, there's three different targets. There is desirability, I want the smallest possible value. So, if you're buying a house, it could be, I want the lowest possible electric bills where zero is the goal. It's not gonna be zero, but I'm looking, of all the ones that pass the must, now I'm looking at all the houses, and I'm saying "I want the lowest possible electric bill". That's a Smaller-is-Best. 0:13:35.9 BB: Larger-is-Best is I want something which is as big as possible. It could be I want the most roof facing the sun, in case I put solar in. That's a Larger-is-Best characteristic, where Taguchi would say the ideal is infinity, but the bigger, the better, as opposed to Smaller-is-Better. And, the other characteristic is what Taguchi calls Nominal-is-Best, is I have an ideal single value in mind. And in each case, the reason I point that out is that desirability is about going past acceptability and saying amongst all the things that are acceptable, I want the smallest, I want the largest, or I want this. It is a preference for one of those. 0:14:19.4 BB: So, I thought... I was using that to explain to this friend the other day, and I thought that would be nice to tie in here. That desirability is a focus on of all the things that meet requirements, now I want to go one step further. That's just not enough. All right, so now let's get into Paradigms B and C. And I want to use an exercise we used in the first series. And, the idea for our audience is imagine a quality characteristic having a lower requirement, a minimum, otherwise known as the lower spec, the lower tolerance. So, there's a minimum value, and then there's a maximum value. And, when I do this in my classes, I say "let's say the quality characteristic is the outer diameter of a tube." And, then so what I'd like the audience to appreciate is we've got a min and a max. 0:15:18.9 BB: And, then imagine your job as listener is to make the decision as to who to buy from. And. let's say we've got two suppliers that are ready to provide us with their product, these tubes that we're gonna buy. And, your job as a listener is to make the decision as to who to buy from. Who are we going to buy from? And, so we go off and we tell them, "Here's the min, here's the max," and they come back. And, they each give us a distribution. And, so what I'd like the audience to think about is a distribution. Just think very simply of two normal distributions, two Gaussian distributions. And, let's say the first distribution goes all the way from the min to the max. It takes up the entire range. 0:16:08.5 AS: So wide and flat. 0:16:12.1 BB: Wide and flat. That's supplier one. And supplier two, let's say is maybe three quarters of the way over. It's incredibly uniform. It uses a very small fraction of the tolerance. So that's tall and narrow. That's distribution two as opposed to wide and flat. So, imagine we've got those two to buy from. But imagine also, and this is a highly idealized scenario. And, I use this and this is why I want to share it with our audience. Because it becomes a great way of diving into what I think is a lot of confusion about meeting requirements. And, so what I want you to imagine is that no matter who you buy from, they both promise that they will deliver at the same price per tube. 0:17:00.8 BB: So, no matter who you buy from, price-wise, they are identical. To which I'd say that's highly idealized, but that's a given. Criteria number two, the delivery rates are the same. So, we cannot differentiate on delivery. We cannot differentiate on price. The third condition we find out is that everything they deliver meets requirements, 100%. So, if there is any scrap and rework, they don't ship that to us. So, everything they deliver meets requirements. And, again, that's highly idealized. 0:17:41.6 BB: Number four is the distributions are in control. And, that means that the processes are predictable and stable. And, that's guaranteed. So, imagine these distributions day by day every order is the same shape, the same average, the same amount of variation. Also, it will never change. It will never change. And, the other thing I want to point out in this fourth point here is that your job as the buyer is to buy these. They are used as is within our organization. , 0:18:15.5 BB: And, the fifth point is that there's a min and a max. And, so I've been using this exercise for, gosh, going back to 1995, and I throw it out there and then I show them the distributions. I say "same price, same schedule, delivery rate, everything meets requirements, distributions never change shape or location. You're going to use as is. And there's the min, there's the max. Who do you buy from?" And, I give people not only do we buy from one or two, but I also say I'll give you a third option. 0:18:51.5 BB: The third option is it doesn't matter. It doesn't matter. So, what I find is that three quarters of the audience will take distribution two, the narrow one. And when I ask them, why do you like distribution two? They say, "because it has less variation". I then say, "From what?" Then they say, "From each other." And, that's what a standard deviation is, variation from each other. So roughly 75% plus and minus... [overlapping conversation] 0:19:25.8 AS: When you say of each other, you're talking about each other curve or each other item in the... 0:19:31.3 BB: Each other tube. So, the amount of variation from all the tubes are close together, so the variation from each other. 0:19:38.6 AS: Okay. Each item. Yeah, okay. 0:19:41.8 BB: Standard deviation is the average variation from the average value. So, when I ask them, why do you like two? Okay, and then I asked the ones who take the wide one in the middle, I say, "why do you like that one," and they say because... And, actually, we'll come back to that. This is pretty funny. They will take that, but a very small percent say it doesn't matter, and here's what's interesting, if I didn't show the distributions, if all I did was say there's two suppliers out there, the same price, same schedule, that guarantee zero defects, the results will never change. Here's the min, here's the max, I'm willing to bet if I didn't show the distributions, people would say "it doesn't matter, I'll take either one". But, as soon as I show them the distributions, they want the narrow one. And, I use this for our attendees, this is a great way to show people that they really don't believe in tolerances, 'cause as soon as you go past meeting requirements, what you're really saying is, there's a higher bar. 0:21:05.6 AS: Okay, so requirements would be... Or, tolerances would be the extremes of that flat, wide curve. And, any one of those outcomes meets the tolerance. 0:21:17.5 BB: Yes, and so for companies that are striving to meet requirements, why is it when I give you two distributions that meet requirements... Why is it when I show you the distributions, and I'm willing to bet if I don't show you the distributions and all you know is they're 100% good, then you say "well, it doesn't matter," Well then what changes when I show you the distributions? 0:21:43.6 AS: I know why I'd choose the narrow one. 0:21:48.1 BB: Go ahead. 0:21:49.1 AS: I know how damn hard it is to reduce variation and I forget about any tolerance of anything, if I have two companies that show me a wide distribution, and another one shows me a narrow one, and let's say it's accurate. I'm much more impressed with how a company can do the same exact output as another company, the same product that they're trying to deliver, but they are producing a much more narrow range of outcome, which could be that they just have automation in their production line and the other one has manual. 0:22:27.4 BB: And, I have seen that within Rocketdyne, I've seen processes do that. I have seen the wide become the narrow through automation. Yeah. Okay, so hold that thought then. So, what I do in my graduate classes is I show that... Not only do I give them two options, I give them four options. So, I throw in two other distributions, but really what it comes down to is the wide one versus the narrow one, and then the other two, I throw in there that usually aren't taken, they're distractions. All right, so what I'll do in a graduate class in quality management is to show that and get the results I just showed. If I present the same exercise and then say, "imagine the average value of distribution one, the middle of distribution one, imagine that is the ideal value". 0:23:24.7 AS: That, you're talking about the wide and flat. 0:23:28.4 BB: Yes. So, all I do is I go back to the entire exercise and now I add in a line at the average of the wide distribution, and then go through and ask one more time, who would you take. 0:23:46.3 AS: So, now the dilemma that the listener has is that now we have a, within limits, within tolerances, we have a wide but flat distribution that's centered on the middle point between the upper and lower tolerance. 0:24:06.4 BB: Yeah, yes. 0:24:08.8 AS: And, then we have... Go ahead. 0:24:11.7 BB: Well, yeah, that is distribution one, same as the first part, we went through this, and all I'm doing now is saying, "imagine the average value of the middle is said to be the ideal value". 0:24:29.4 AS: And, now you're gonna tell us that the narrow one is not on that central or ideal value. 0:24:36.2 BB: No, that is still where it is at the three-quarter point, all I've done is now said, this is desirability. I'm now saying "that is the ideal value, that is the target, that is the value we prefer". And, people still take the narrowest distribution number two. 0:24:58.8 AS: I wouldn't take the narrow one because I would think that the company would have to prove to me that they can shift that narrow curve. 0:25:06.6 BB: Well, okay, and I'm glad you brought that up because according to the explanation I gave of equal price, equal schedule, meets requirements. I deliberately put in the criteria that you have to use them as is. So, now I'm forcing people to choose between the narrowest one over there at the three-quarter point, and the wide one on target. And, there's no doubt if I gave them the option of taking the narrow distribution and sliding it over, they would. Every single person would do that. But, when I give you a choice of, okay, now what? So, two things here, one is, is it calling out the ideal of value, 'cause desirability is not just beyond acceptability, it is saying, "I desire this value, I want this parking spot, I want this apple, I want this value". And, that's something we've been alluding to earlier, but that's what I wanna call out today is that... 0:26:13.7 BB: So, in other words, when I presented the exercise of the two distributions, without calling out what's desirable, all I'm doing is saying they're both acceptable, which do you prefer? But, instead of saying it doesn't matter, I'd like the narrowest one, and it may well be what people are doing is exactly what you're saying is the narrowest one seems better and easily could be for what you explained. 0:26:40.8 BB: But, what's interesting is, even when I call out what's desirable as the value, people will take the narrowest distribution, and so now what I wanna add to our prior conversation is Paradigm A, acceptability, the Paradigm A response would be, it doesn't matter. Choosing the narrowest one, otherwise known as precision, we're very precisely hitting that value, small standard deviation, that's what I refer to as Paradigm B, piece-to -piece consistency. Paradigm C is desirability being on the ideal value, that's piece-to-target consistency. And, in Dr. Taguchi's work, what he's talking about is the impact downstream of not just looking at the tubes, but when you look at how the tubes are inserted into a hole, perhaps, then what he's saying is that the reason you would call out the desirable value is what you're saying is how this tube integrates in a bigger system matters, which is why I want this value. 0:27:54.2 AS: Okay, so let's go back, A, meet requirements, that's acceptability. Anything within those tolerances we can accept. B is a narrow distribution, what you called precision or piece-to -piece consistency. And what was C? 0:28:12.8 BB: C is, I'll take the wide distribution where the average value is on target, that's piece- to-target consistency. Otherwise known as accuracy. 0:28:27.3 AS: Okay. Target consistency, otherwise known as accuracy. All right, and then precision around D is precision around the ideal value. 0:28:37.7 BB: Well, for those that want to take the narrowest one and slide it over, what you're now doing is saying, "I'm gonna start with precision, and I'm going to focus on the ideal". Now, what you're doing is saying, "step one is precision, step two is accuracy". 0:28:56.4 AS: Okay. And step three or D? 0:29:00.9 BB: Paradigm D? 0:29:02.7 AS: Yeah. 0:29:02.7 BB: Is that what you're... Yeah. Paradigm D would be the ability to produce, to move the distribution as needed to different locations. 0:29:17.4 AS: The narrow distribution? 0:29:18.9 BB: Yes, and so I'll give you an example in terms of, let's say tennis, Paradigm A in tennis is just to get the ball across the net. I just wanna get it somewhere on the other side of the court, right. Now that may be okay if you and I are neighbors, but that doesn't get us into professional level. Paradigm B, is I can hit it consistently to one place on your side of the court. Now, I can't control that location, but boy, I can get that location every single time. Next thing you know, you know exactly where the ball is going, and that's Paradigm B. Paradigm C is I can move it to where I want it to go, which you will eventually figure out, so I can control where it goes. Paradigm D is I can consistently hit any side of the court on the fly. 0:30:11.0 BB: So, Paradigm D is I can take that narrow distribution and move it around for different customers, different applications, and Dr. Taguchi refers to that as Technology Development, and what Taguchi is talking about is developing a technology which has incredible precision in providing your sales people the ability to move the next move it to accuracy and to sell that product by tuning it to different customers as you would in sports, move the ball around to the other side of the court. So now you're going to the point that you've got incredible precision, and now you've got "on demand accuracy," that's Paradigm D. Paradigm C is I can do one-size-fits-all which is, which may be all you need for the application. 0:31:06.9 AS: I wanna separate the Paradigm B, the narrow distribution and that's precision around some value versus Paradigm D is precision around the ideal value. 0:31:20.7 BB: And, the idea is that desirability is about an ideal value. And, so if we're talking piece-to-piece consistency, that means it's uniform, but I'm not paying attention to... I have a value in mind that I want. And that's the difference between Dr. Taguchi's work, I mean, it's the ability to be precise. Again, accuracy, desirability is I have an ideal value in mind. And acceptability is it doesn't really matter. Precision is uniformity without accuracy. And so, if you are... What Dr. Taguchi is talking about is, is depending on how what you're delivering integrates, being consistent may cause the person downstream to consistently need a hammer to get the tube into the hole. 0:32:24.2 BB: So, it's consistent, but what you're now saying, what Taguchi is saying is, if you pay attention to where you are within requirement, which is desirability, then you can improve integration. And, that is my explanation for why Toyota's products have incredibly reliability, that they are focusing on integration, not just uniformity and precision by itself. 0:32:49.8 AS: I would love to put this in the context of a dart thrower. The Paradigm A meeting requirability or acceptability, they stand way behind and they throw and they hit the overall dart board. 0:33:04.3 BB: Dart board. It's on the board. Yes. 0:33:07.2 AS: And, the narrow distribution is, well, they hit the same spot over to the left, right towards the edge, they hit that spot consistently. And, then basically, I'm gonna jump to D just because I'm imagining that I'm just gonna ask the guy, Hey, can you just move over just a little bit, and I'mma move them over about a half a foot, and when I do, you're gonna start throwing that dart right at the same location, but over to the right, meaning at the target. The center of the dart... 0:33:43.9 BB: The bull's eye. Yeah. Yeah, well, that's... And you call that C or D? 0:33:47.6 AS: I call that D. 0:33:49.5 BB: No, I would say, let's call that C being on target, meaning that C is, for games of darts where the most points are being on the bull's eye, that's Paradigm C. 0:34:04.0 AS: So accuracy, yeah. 0:34:05.4 BB: Paradigm D would be a game in which the ideal value changes. So now, okay, now I watch the... When I play darts, I'm sure there's lots of darts games, but one game we used to play it in our cellar at home was baseball. So, the dart board is divided into has numbers one, two, three through, and you'd go to... There'd be a wedge number one, a wedge number two, a wedge number three, that's Paradigm D that I could hit the different wedges on demand. But that's what it is. So A is anywhere in. B is consistent, precision, but again, the idea is if you can move that, but now what we're talking about is, is there an impulse to move it or are we happy just being precise? What Taguchi's talking about is the value proposition of desirability is to take precision, take that uniformity and move it to the ideal value, and what you've just done and doing so, you're now focusing on not this characteristic in isolation, you're now focusing on how this characteristic meshes with another characteristic. And, it's not just one thing in isolation, one thing in isolation does not give you a highly reliable automobile. 0:35:38.9 AS: Is there anything you wanna add to that, or are you ready to sum it up? 0:35:45.0 BB: No, that's it. The big summation is, we've been building up to the contrast between acceptable and desirable. I just wanted to add some more fidelity. Desirable is I have a value in mind, which Dr. Taguchi referred to as a target. So, for people at home, in the kitchen, the target value could be exactly one cup of flour. We talked earlier about our daughter, when she worked in a coffee shop and then, and at home she'd give us these recipes for making coffee and it'd be dad, exactly this amount of coffee and exactly that. And, we had a scale, it wasn't just anywhere between. She'd say "dad, you have to get a scale." I mean she was... We started calling her the coffee snob, 'cause it was very, this amount, this amount. So, in the kitchen then it's about precisely one cup. Precisely one this. And that's desirability. 0:36:40.6 AS: And, I was just thinking, the best word for that is bull's eye! 0:36:48.3 BB: Yes. 0:36:48.8 AS: You hit it right on the spot. 0:36:50.6 BB: Yeah. 0:36:51.6 AS: Great. Well, Bill, on behalf of everyone at The Deming Institute, I wanna thank you again for this discussion. It was not only acceptable, it was desirable. For listeners, remember to go to deming.org to continue your journey. And, if you want to keep in touch with Bill, just find him on LinkedIn. He'll reply. This is your host, Andrew Stotz, and I leave you with one of my favorite quotes from Dr. Deming, "people are entitled to joy in work."