Dec 5, 2023
Should we strive to better understand what happens "downstream" to our defect-free work? No matter the setting, if our work meets requirements and we pass it on, are we responsible for how well it integrates into a bigger system? In this episode, Bill Bellows and Andrew Stotz expand on the interaction between variation and systems and why Dr. Deming regarded Genichi Taguchi’s Quality Loss Function as “a better description of the world.”
TRANSCRIPT
0:00:02.8 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 am 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 today is, in episode 13, Integration Excellence, part two. Bill, take it away.
0:00:31.4 Bill Bellows: Thank you, Andrew. Always a pleasure to connect with you. Alright.
0:00:40.1 AS: Mine too.
0:00:40.1 BB: [laughter] In episode 12, I thought it was great. We shared perspectives on the human side of integration, what it means to be connected, to be synchronous, to feel included, to feel connected, to feel included or connected when something good happens where you're like, well, I was part of that, or to feel separated is when something bad happens. And, we somehow have the ability to not feel associated with that. I pass the puck to you and you hit the slapshot, it goes into the stands, off the goalkeeper. Y'know, girl gets hit in the head and you feel bad, but I go home and I can sleep. And so why is that? And so anyway, but I thought, and listening to it, and I thought it was a lot of fun to look at the human side of feeling connected or feeling separated. And what I wanted to get into tonight, and perhaps in another episode as well, is the physical side of connections.
0:01:46.5 BB: One thing I wanted, and I got a couple anecdotes. I had a woman in class at Rocketdyne years ago, and she said, "Bill, in our organization, we have compassion for one another." And I said, "Compassion is not enough." And, and so you, Andrew, could be in final assembly at this Ford plant, where you're banging things together with a rubber mallet 'cause they're not quite snap fit, and you're banging them together. I mean they all meet print perhaps, but where they are within the requirements is all over the place, and you're having to bring them together. That's called integration. And so when this woman said, in our organization we have compassion for one another, I said, well, that's like me saying, "Andrew, I feel really bad that you're, I can't believe, Andrew go home. You can bang that together tomorrow. You've been banging it together all day." And what I said to her is that "compassion is not enough."
0:02:54.7 BB: When I feel connected to what you're doing, when I begin to understand that the parts you're banging together meet requirements, but how they meet requirements is causing you the issue. Now, the compassion plus my sense of connection, now we're talking. But short of that, what I think is we have organizations where as she would say, we might feel bad for others. And it means I hear about your injuries and your ergonomic training because of all this, but I don't, until I feel associated with that, I just feel bad. But feeling bad is not enough. But I like that, that sentiment. But what I wanna look at tonight is a greater sense of Dr. Taguchi's so called Loss Function and look at more why we should feel more connected to what's happening downstream. So I wanted to throw that out. [chuckle] On the topic of variation, I just started a new cohort with Cal State Northridge University. And this is my, fifth year in the program doing an eight week class in, seminar in quality management. And the cohort model is, anywhere between two dozen and 30 some students that start, the ones I'm getting started a year ago.
0:04:23.5 BB: And they have class after class after class after class. Then a year into the program they get to meet for eight weeks so then onto other professors in the program. So I was showing them, first quarter, second quarter data points from an incident that happened at Rocketdyne years ago. And I was in a staff meeting and the vertical axis is number of accidents per employee. And the horizontal axis is quarter one, quarter two. So the quarter one data point is there, and I don't have the original data, the original data doesn't matter. But what I say to the students is, imagine we've got the first quarter data, what would you expect for the second quarter data? And what's funny is a number of them said, it should be lower. And I said, "Well, based on what?" And it's like, said "Well, we're gonna go off and study what went wrong and we're gonna improve the process."
0:05:20.6 BB: And I said, "Okay, that's all right." So then, I said, "I'll accept that, that's a possibility." Well, then I showed them the actual second data point was lower than the first, which in the meeting I was in, led to the question from one of the senior managers to one of the more, let's say the vice president of operations, "Hey, Andrew, why is safety improved?" To which the executive said, "Because we've let them know safety is important." And so I asked him, "So what do you hear in that?" And we went around and we went around and we went around. It's not the only time it has happened that what they're not hearing is the separation that "we" have let "them" this this. And so in part, I think with my Deming perspective finely tuned. I pick up on those things. And they're not picking on it picking up on it yet which is which is fine. And then but I kept asking, kept asking, kept asking. And then one person said, "Well maybe we need to look for a pattern." I said, "Oh brilliant. What if we've got this run chart of all this extra data?" So then I got them to buy into how easy it is to take two data points draw conclusion up and down. That's called variation. And so it was neat to... The first conversation with them on the topic of variation was really cool. And there's so much more to follow. Well then it, what I wanted to follow with this once upon a time our son when he was in third grade this is 20-some years ago invited me to come to his class.
0:07:07.6 BB: And I don't recall why other than he said, Can you come talk to the class? And I said, Okay fine. So my biggest concern was that the teacher wouldn't know I was coming but she knew I was coming so it was good. So I walk in talked with her briefly and I said I've got some things I'd like to do. She's like oh, I didn't wanna monopolize. But she said okay why don't you show your video? I said I got a video of rocket engines blah blah blah. And then I've got a little exercise I wanna do. Okay we'll do the video then we'll do some reading. So we're doing the reading. And so I'm helping her with the reading. And then what I noticed is now and then a word would come up and she'd write the word on the whiteboard and ask the students if they understood the word. So I clued in, I cued in on that. So when it got to me I wrote the word theory on the whiteboard. This is third graders Andrew, third graders. [chuckle] And I said do any of you know what a theory is? And a one of the girls Shelby whose name I'll never forget, she raises her hand and she says a theory is a prediction of the future. Third grade Andrew third grade! [chuckle] right? Now...
0:08:18.8 AS: And you know what they'd say now they'd say Ethereum is a type of cryptocurrency. [chuckle] Oh Ethereum. No no "theory" not "Ethereum." [laughter]
0:08:30.2 BB: You're right. You're right.
0:08:31.6 AS: Okay. That's a great answer.
0:08:33.9 BB: Well oh but what I tell my students is I didn't correct her. I didn't say well technically a theory is a prediction of the future with a chance of being wrong. But we'll just, I just, oh we'll just stop with that. So I invited her to the front of the room. So she comes to the front of the room and I brought with me this little plastic bag with half a dozen marbles in it. And the bag was also a holes from a three hole punch, little dots of paper. So I held the marble up and I said Shelby I'm going to drop the marble from this height predict where it will land. And what I tell students is she was able to predict where it would land without any data.
0:09:18.3 BB: So she predicts the first data point, the marble lands someplace else. I marked the spot with a marble. I then said okay Shelby I'm gonna drop it the second time. Where will it land? And I'll ask people in class so where do you think she predicted, exactly the same spot of the first drop [chuckle] Exactly right. That's what we do as adults. And so we went through this cycle again and again. And and finally after about 10 drops where these you know 10 different dots on the floor I said Shelby where's it gonna land? And she drew a circle, she said somewhere in here which is kind of like a control limit you know kind of thing.
0:09:55.9 BB: So the one thing I'll say is and I'm sure you've heard people say well you can't predict the future. No, as Dr. Deming would say [chuckle] you know he gave the example you might recall of how will I go home? I'm gonna take a bus. Will the bus... I'm gonna take the train. Will the train arrive? And so I'd ask adults in the class that says how many drove here today? All the hands go up. And I said so at the end of the day will you walk in the direction of where you left your car? Yes. What is your theory? It's still there. [chuckle] Is that a guarantee? No! [chuckle] So I throw that out as a predictions and her sense of variation and this sense of a third grader not acknowledging, I mean one understanding having some sense of a theory, not a lot of understanding of variation but I don't think that's unique to third graders.
0:10:51.8 BB: So that brings us to...there's variation. We can look at the variation in the Red Beads. Okay the Red Beads are caused by the system not the workers taken separately. Then we got into variation and things that are good. And when I introduced the students to last night in class is, I asked them "So how often do you go to meetings where you work to discuss things that are good and going well?" And I get the standard answer, "rarely." I said, "Well, why is that?" "Well 'cause we got, we're focusing on the bad." They said, "to make it good." "Well why do we focus on the bad to make it good? Why don't we focus on the good?" "Well the good is good." And we went around the room, went around the room online and and I said "what's the likelihood that we could prevent bad from happening by focusing on the good while it's good?" And it's like, "...interesting." And so where that leads us to is, is two aspects of looking at things that are good.
0:11:57.1 BB: One is the better we understand the variation of things that are good whether that's on a run chart or a control chart. My theory is we could prevent bad from happening by keeping track of the bad. Whether it's your pulse, your weight, [chuckle] how much gas is in your car. And so there's if we focus, if we pay attention to the good with some frequency you know every second, every hour, once a month, whatever it is, we could prevent an accumulation of damage to an appliance at home. Another aspect to focusing on things that are good is that it can improve integration which is boom, here we are. And that integration that I mentioned last time that understanding integration could be looking at candidates for a new hire and looking for who is the best fit because there's degrees of fit. Fit is not absolute. Last time we talked about reflections of an engineer who is worried that his hardware on the space shuttle main engine may have contributed to the disaster of the second... Of the Columbia space shuttle blowing up in reentry. Well let me share another story from a coworker at Rocketdyne.
0:13:19.8 BB: And this guy's father worked at Rocketdyne in the '60s. So in 1999, 30 years after the lunar landing, there's news teams, you know, from the local TV stations and television. It's 30 year anniversary of the Lunar Landing. And Rocketdyne was known for the Apollo engines that get the vehicle off the ground, as well as the engines that got the, Orbiter off the moon. So there's an article in the newspaper a couple days later, and this coworker is quoted and he says, "Boy, I would've loved... My father worked here back in the '60s, just to be a fly on the wall would be so cool. Oh my gosh, it'd be so cool." And the article ends with him saying how exciting it is to feel like you're part of something big. That's what we talked about last time.
0:14:09.8 BB: And I used to use that quote from him on a regular basis because it, the article was about something that happened at Rocketdyne. Then I would share that this is a quote from a coworker. And after quoting him for several years, it dawned on me, I've never met this guy, so I call him up one day and he answers and I say, "Hi, this Bill Bellows." And he laughs a little bit. And I said, "have we ever met?" And he says, "No, no, no," he said, "But you quote me in your class." And I said, "Well, I apologize for never calling you sooner." I said, "I do quote you." And I said, "Let me share with you the quote." I said, "you feel how exciting it is to feel like you're part of something big?" To which he says, "I wish I still felt that way." [chuckle] And I said, "can I quote you on that?" And so you can join an organization with this sense of being connected, but then depending on how the organization is running and you're blamed for the Red Beads, that you may lose that feeling.
0:15:15.6 BB: And on another anecdote, it's pretty cool. Our daughter, when she was in fourth grade, was in a class, they were studying water systems. And the class assignment was to look at a, they had an eight and a half by 11 sheet of paper with a picture of a kitchen sink on it, like a 3D view of a sink with a pipe out and a pipe in. And the assignment was, we're about to study water systems. How does the water get to the sink, where's the water go?
0:15:47.2 BB: And so my wife and I were there for the open house and there were 20 of these on the wall colored with crayons showing all these different interpretations of water coming in, water going out. And I was fascinated by that. And eventually got copies of them and the teacher wasn't sure what I was doing with them. Well, I turned them into laminated posters. And so I gave one to our daughter one day. I said, take this to Mrs. Howe so she sees what we're doing. And so the following weekend I bumped into this woman at a soccer field, but she wasn't dressed like a teacher. She's dressed in a hoodie. And she says to me, "Allison shared with me the posters." And I'm looking at her thinking, "how do I know who you are?" She pulls the hood back. She says "I'm Allison's fourth grade..." Oh! I, her comment was when Allison shared with me how you're using those posters, handing them out, and people are inspired by them. And she says, "I cried." So that you get that emotion for free Andrew. [chuckle] Right. And that's all the integration stuff.
0:16:58.5 BB: Now let's talk about Dr. Taguchi and his Loss Function. So, um, the Taguchi Loss Function says Dr. Deming in Out of the Crisis is a better view of the world. The Taguchi Loss Function is a better view of the world. Dr. Taguchi says following...
0:17:15.3 AS: Wait a minute. I was confused on that. You're saying Deming is saying that Taguchi is better, or Taguchi is saying Deming's better?
0:17:22.3 BB: Dr. Deming in The New... In Out of the Crisis, Dr. Deming wrote "the Taguchi Loss Function is a better view of the world."
0:17:30.3 AS: Okay, got it.
0:17:34.5 BB: And that's what amongst the things that I read into Deming's work and I thought, boy, that's quite an endorsement. Dr. Taguchi is known for saying quality is the minimum of loss imparted to society, to the society by a product after shipping to the customer. So what does that mean? And we'll come back to that. Deming met Dr. Taguchi in the 1950s. There's a, at least once, there's photos I've seen in Deming's archives of the two of them on stage at a big statistical conference in India, and I know they met in September, 1960 at the Deming Prize ceremony where Dr. Taguchi was honored with what's known as the Deming Prize in Literature. There's Deming prizes for corporations, and there's also Deming prizes for individuals.
0:18:35.0 BB: And Taguchi won it 1960 for his work on the, on his, this quality-loss function concept. 1960. So then in 1983, Larry Sullivan, a Ford executive, was on a study mission to Japan, and he wrote an article about this for the American Society for Quality in 1983 the title of the article is “Variability Reduction: A New Approach to Quality,” so if any of our listeners are ASQ members, well I'm sure you can find a copy of it. The Variability Reduction: A New Approach to Quality. Well, Andrew in 1983, Sullivan's article, 23 years after Taguchi's awarded this Deming Prize in literature, I'm convinced that's the first time Taguchi's Loss Function was heard about in the States. 23 years later. And in this article, Sullivan says, he says, "In March of 1982, I was part of a group from Ford that visited Japan, we studied quality systems out of variety of suppliers," this is ostensibly the first time the auto industry in the States is sending people to Japan.
0:19:52.8 BB: Right so 1980, summer of 1980 is the Deming documentary Why Japan? If Japan Can, Why Can't We? And so here Ford is in 1982, sending a team over. I know it was the late '80s, I believe, when Boeing sent executives over. So then in this article, he says, "The most important thing we learned, right, in this study mission, is that quality in these companies means something different than what it means in the US. That it's a totally different discipline." And so this is like the beginnings of people hearing about Dr. Deming in 1980. They're now hearing about Dr. Taguchi's work through Larry Sullivan. And it turns out Larry Sullivan and Dr. Taguchi became business partners and set up Dr Taguchi's consulting company in the States, which still exists. So they became fast friends and I've met the two of them many times.
0:20:53.6 BB: What Taguchi is saying is, is when it comes to things coming together, we talked about integration, whether that's combining, mixing, joining, weaving, this is the synchronicity. So in sports, we're talking about not, not where I am on the field, but where I am relative to the others, in music, and we're talking earlier about music and I've, I've played a musical instrument one time, Andrew with a group and I was with a, hockey band on a road trip when I was in college. And the cymbal player, they were missing, so they asked me to bang the cymbal, "you want me to do what?"
0:21:36.9 AS: When we signal you.
0:21:39.4 BB: So I'm boom! and what I didn't realize is I'm controlling the pace, like being in is like, okay, slow down, slow down. And I and a former student last year in the Cal State Northridge class who plays with one of the Beach Boys, and I went to watch her in the play and I was asking about these speakers, which are on stage, facing the players. And I said, so what are those about? She said, "Those help us stay synchronized." I said, "what do you mean?" She says, "the speakers next to me," she's the keyboard player. She said, "What I'm listening to in those speakers is the drumbeat. I need to make sure that I am playing synchronous with a drummer." And then what about the others? "Well, the others have their own speakers synchronized. They get to select who they wanna be synchronized to." And so I throw that out because we take for granted when we're listening to Coldplay, whoever these musicians are, we're not paying attention, at least I'm not paying attention to what if they're playing it... What if they're not as synchronous? How would that sound? 'Cause we're so used to it sounding pretty good.
0:23:00.1 BB: And, um, so there we go with synchronization and things fitting together, it's not just that the note was good, but is it played at the right rhythm and pace and, um, you know, with timing. So we talked about the Loss Function. We talked about last time about ripeness of fruit. Depending on what we're doing with the bananas, we wanna put it into a muffin mixed or eat, slice it up. Are we looking for something soft and hard? And I say that because what Dr. Taguchi is talking about is for a set of requirements, a min and a max, we're used to a sense of anything between the min and the max is okay, is "good."
0:23:45.2 BB: What Taguchi is saying is there's the possibility that there's an ideal place to be. And how do you know what that ideal place to be is? Well pay it, as you're delivering that piece of fruit to the next person, whatever it is, to the next person, deliver them something on the very low end of the requirement and see what they do with it. Then, it could be the next hour or the next time you give them something a little bit, a little bit further along that axis. How are they doing? How are they doing? How are they doing? And what you're looking to see is, how, how does, what is the effect of where you are within requirements on them? And this is how Toyota ends up with things being snap fit, because they're not just saying, "Throw everything to Andrew in final assembly." They all come together.
0:24:42.3 BB: My theory is they're doing what we do at home, at home I create the part, I cut the piece of wood. I'm, making the part, but I'm also using it. So I'm the one responsible for the part and integration, in a work setting that may not be the case. So what Taguchi is talking about is there could be a sweet spot in the requirement. And so towards that end, if we're talking about baseball in a strike zone, the World Series is teams are defined, not that I was gonna watch this year, the Dodgers, we're out of it. But in baseball, there's, for those understand baseball, there's a strike zone. If the ball somewhere in that rectangular ball zone is called a strike, outside is called a ball. And depending on who the batter is, it might not matter where the ball is in the strike zone, 'cause this player can't hit the ball anyway. But for another player, you may have to put that strike somewhere in particular to make it harder for them to hit. And that's what the loss function is about, is, is paying attention to how this is used and I wanna share a couple of stories that are, one that's kind of hard to believe. Well, I'd say one that's easy to believe. As you're driving down the highway, Andrew, in Los Angeles, right? You've lived out here.
0:26:07.2 AS: Oh, yeah.
0:26:07.4 BB: And no matter where you're driving down, right, do you stay to the left side of the lane, Andrew? Do you stay to the right side of the lane? Or do you kind of go down the middle of the lane, Andrew?
0:26:17.9 AS: I'm kind of middle of the lane guy.
0:26:20.5 BB: Yeah. And I think that people in the other lanes, you know, like that 'cause I know when I drift to the left, you're like, Hey, what are you doing? So being towards the middle is saying, I get the entire length of myself, but being down the middle is probably, what is that? It's minimum loss to myself and others. So I spoke at a, at a NASA conference ages ago and learned, this is uh '97, '98 timeframe, and I learned that the two greatest opportunities for destruction of the space shuttle are at launch, you can have a catastrophic failure, or at landing. And so at launch, it could be a problem with the engine, any of the engines or the solid rocket motors. Okay, so that I can understand. But I'm thinking, what's the issue with landing? Well, I say, well, the issue with landing at that time was the space shuttle's coming in at a couple hundred miles an hour.
0:27:24.9 BB: And when you're landing on a dry lake bed called Edwards Air Force Base, it's not a big deal. You got all that open space anywhere you want. You just get her down. But then in that timeframe, NASA converted. It was easier for them to have the shuttle land in Florida because they don't, they don't have to fly the shuttle across country. The shuttle is going to land there, launch there. So what they were talking about is, a lot of the pilots for the space shuttle are military pilots. They're used to landing in the center of the runway, Andrew, in the center of the runway. Why? 'cause they're landing on an aircraft carrier. And if I'm a little bit too far from the center, one way or the other, I either crash into the structure or I'm in the ditch and enter the water. So they've got these military pilots landing the space shuttle, wanting to be right down the center. And so they said what happened was if they land and they're a few feet to the left or to the right, going a couple hundred miles an hour, should they quickly steer the nose gear to be on the center?
0:28:32.5 BB: And he said, when you're going that fast, if you steer, you may cause the shuttle to just flip. When you're, once you touch down, don't steer to the center of the runway. Just go, go straight. No more steering. And they kept having this message and it kept being ignored and they kept having the message that kept being ignored so what was the solution, Andrew? You ready?
0:28:58.7 AS: Yes, here, tell me.
0:29:00.8 BB: They painted the center stripe to be wider. [laughter]
0:29:05.5 AS: I was thinking they were going to paint like 10 stripes so that there was no center one.
0:29:10.3 BB: So the center stripe is like three feet wide. You can't miss it. Well, and so I use that because what they're saying is when you land at the Kennedy Space Center, you could be off target left and right a lot, and it's not a big deal, we got a lot of space here.
0:29:29.4 AS: Yep.
0:29:29.6 BB: And what does that mean relative to loss of the vehicle, relative to bad things happening downstream? The loss function that Dr. Taguchi would describe as a parabola, and a parabola being a curve that has a minimum, and then the curve goes up faster and faster to the left, faster and faster to the right. That's if the parabola opens up, it could open down. But in this case, Taguchi draws the loss function as being opening upwards as like a bell and it gets steeper and steeper. But, what, but depending on your system, it could be very steep, which is you're landing on an aircraft carrier, or it could be very shallow.
0:30:13.6 BB: So when I ride on a bike trail in Santa Clarita where I live, I go down the middle of the bike trail. And to my right, depending on which direction I'm going is a split rail fence so I don't go into the Arroyo, which is this gully for all the water running off. And so there's... I go down there and the worst, I stay away from that split rail. When I ride in Long Beach where you went to college where our daughter lives, there is no split rail. So I stay not in the center when I ride in Long Beach. I ride to not the center of my lane, I steer closer to the to the center of the overall lane, which means I'm closer to the bikes going the other way. And that's and that's my understanding of: I go off that off that side is gonna be a bad day.
0:31:08.0 BB: And so that's what Taguchi is saying relative to the loss function. But I think a better way to think about loss, I think that may be kind of a weird concept. I think if we think about integration, and in making the integration easier or harder. So again, if we're talking about space shuttle landing, maybe the loss makes sense. But if we're talking about putting things together, we've talked about the snap-fit that Toyota pickup truck that Toyota was producing in the late 1960s. And what struck me when I first read that is, Holy cow, they've developed a system of hardware which goes together without mallets, and I immediately associated that with what I had heard that Dr. Taguchi was influencing, working with them, consulting with them back in the '50s. And I thought that kind of fits. And so why aren't things here in the States, why are they being banged together? Because over in the States, going back to Larry Sullivan's article, we've got an explanation of quality which is "part" focused. Everything meets requirements. And so what really amazed me is that Toyota in the late '60s, had things which were going together well.
0:32:25.9 BB: Ford in 1982/83 timeframe, they had been working with Dr. Deming for a couple years. They discovered that a transmission they had designed and were building was also being built by Mazda. And part because they owned one third of Mazda and they were outsourcing production. And these transmissions went into Ford cars. And what I've mentioned in a previous episode is that the Ford warranty people figured out that the Mazda transmission, which was designed by Ford, but built by Mazda, had one third fewer complaints than the Ford transmission designed by Ford, built by Ford. And in this study that Ford did, led by their executives, and then they sent out the documentation to their supply chain and it, and it talked about the need to... Their explanation was what Mazda was doing was what's known as "piece to piece consistency." And what they found is that the parts, instead of being all over the place in terms of dimensions and whatnot, that they were far more uniform, yet what you won't hear in that video, what they talk about is within Ford, we're all over the place we're consuming the greatest, a big portion of the tolerance. We've got scrap and rework. But these Mazda parts, boy they only consume a fraction of the tolerance compared to us. And that's the difference. And that's the difference.
0:34:02.6 BB: And so what I wanna close with is, having less variation is not the issue that gets us back to precision, but not accuracy. So my explanation is that Mazda was actually focusing on accuracy - being on target of the respective parts. And as a result, they got great functionality outta the transmission. But what Ford, at least, I'm willing to bet the path Ford was going, was saying, "oh look Andrew, their parts are more consistent than ours. Consistency is the name of the game." And that's precision, not accuracy. So what I wanted to do tonight is build upon what we did last time, bring it to this loss function as being a parabola. Depending on what happens downstream, you don't know how steep that parabola is, and not knowing how steep it is, we don't know how much effort we should spend on our end upfront providing those components to improve integration 'cause we don't know how bad the integration is.
0:35:17.6 AS: And that's a wrap. 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. If you want to keep in touch with Bill, you can just find him right there on LinkedIn. This is your host, Andrew Stotz, and I'll leave you with one of my favorite quotes from Dr. Deming. "People are entitled to joy in work."