TQI_S2018_07

Last time, somewhere here, have a pointer. Okay. Last time we were finishing up on what is TQM, and there was this one student's point of view. And a lot of other people think it's just a bunch of buzzwords. But some people — like, anybody know who Larry Bossidy is? He was, I mean, may still be alive, he was the CEO of Honeywell. Before that it was called — anyway, I can't remember what — anyway, I can't remember the name of the company before that, but he was a very kind of well-known CEO of a forty billion dollar company. And he mentioned to Jack Welch what TQM was. Jack Welch adopted it for General Electric in 1995. Ray Stata in 1989 — there was this book American TQM, started the Center for Quality Management in the Boston area, and he was, he's one of the larger donors to MIT, the Stata Center, you know, the ugly building by John Geary, or what's Geary's first name — anyway, the famous architect — Frank Gehry, thanks. Gehry, Frank Gehry, thanks.

So anybody know what the inspiration of the Stata Center is? This is an aside, but I got it directly from the MIT chief architect the day after she had met with Frank Gehry, when they were designing the building. And she was in his conference room in Los Angeles and waiting for him, and he comes in, he starts crumpling up paper and throwing it on the table, and he has this pile of crumpled paper. He says, "That's what your building is gonna look like." And it does. It looks like a pile of junk. Okay, sorry, we all have our opinions of it.

But anyway, is TQM the renaming of hundred-year-old concepts? And I've pointed out a lot of the management concepts in TQM have been around since people like Henry Ford and Sakichi Toyoda and others. Is it worthwhile? Well, some parts of it certainly are. Is it none of the above, or is all the above? It's really all of the above. And that's one of the reasons I wanted to teach this module, is to sort of straighten it out, that it does have a number of good qualities. Bossidy and Welch and Stata are not ignorant men, and they thought it had value. TQM does have some content, as opposed to some other people who talk all the time and the whole world listens to them, but there's almost often no content to what they say. TQM can be considered just like some of these other people and what they say, but in fact it does have some content.

So let's now take today and talk about the methods of TQM. So there's a book that is in its seventh edition, so it's been around for a while. I think it's taught — it's used with people at MIT who teach statistics out of the math department, primarily to people in the Sloan School. D.C. Montgomery, Introduction to Statistical Quality Control. And I'll pass this around. Montgomery has made a fortune writing books that are fairly mathematical, and this — you, it helps to be an MIT student to understand this book, okay. There is some content and some substance to the science behind statistical quality control, and I'm not going to try to teach a course on it. I'm not a statistician. There are courses at MIT on this.

But he talks in the beginning about what quality is. The new American TQM — the first book, I think I had in here, the Ray Stata — Society for American — American Society for Quality Control, so I can't remember the organization. In the eighteen — nineteen eighties, traditionally quality meant fitness for use or fitness to a standard, traditionally. And this is basically what Montgomery says in his Chapter One of that book. But quality is inversely proportional to variability. Okay, this is a modern definition of quality from a statistical point of view: you want to reduce the variability in your manufacturing process. And he talks about some of the metrics: performance, reliability, durability, serviceability. These are three of the "ilities" — they like to call them "ilities" — there's maintainability and others. Aesthetics, features, perceived quality, conformance to standards.

Conformance to standards was one of the old metrics that people used. As long as it met the standard — if General Motors had a standard, you're producing product for General Motors — as long as you met General Motors' standards, they would have no problem with what you're supplying to them. If you didn't, they would threaten to sue you.

Did I tell you the story about the guy who was an automotive supplier in Ohio in the — okay, so another story. I had to give a talk in Columbus, Ohio, to a professional society, this is in the mid-1990s or so. And the head of this professional society ran a manufacturing plant that supplied for GM, Chrysler, Toyota, Honda, all these manufacturing assembly lines in Ohio. I don't even remember what he was supplying, but he made automotive products. And he was late to the dinner, and we all — we went ahead and started the dinner for the professional society without him. He missed the rubber chicken, he got there in time for dessert. And he apologized, and he said that he had gotten a call at eleven o'clock that morning, and he was told by, I think it was Honda, that they wanted him in — was at Marysville, Ohio, whatever — by one o'clock, because they had a problem with the parts he was supplying. And it turns out it's two hours away, so they meant leave right now and come see us at one o'clock, okay. And he did, because they were a major customer.

He got there, and he was late getting back for dinner because he spent the afternoon on this sort of emergency call from Honda. I said, "Well, do you find the Japanese very demanding?" He says, "Oh, they're extremely demanding, but I'd much rather work for them than work for General Motors, Ford or Chrysler." And remember, the time period is 1990s. And I said, "Why is that?" He says, "When they have a problem, they may demand my immediate attention, but when I go there, they're trying to work with me to help solve the problem. If General Motors or Ford has a problem, they will call me up and say, 'Fix it immediately or we're going to sue you.'" Okay. So it's a completely different attitude. This is sort of Douglas McGregor's The Human Side of Enterprise, Theory X, where people are, you know, kind of sitting there, you know, "Fix it or we're gonna sue you," or Theory Y, where people are going to try to work together and solve their common problem. Okay. And the Japanese actually taught the United States in the 1990s that you don't treat people like dirt, okay, you try to treat them with respect.

So in any case, the reason I have this picture of a washing machine here is in the New American TQM, Professor Shiba wrote the book, tells the story after World War Two there was a company that made washing machines, and these are the old-style washing machines in the 1950s that had wringers and stuff. And they were doing fine everywhere in the urban areas, but they kept on having breakdowns in the rural areas. So this is one of the things they ended up saying, well, we want to find out what's going on in the rural areas and why the washing machines are not working when you mount it in the country. So they sent someone out to find what was going on, and they found the problem was the farmers were washing potatoes rather than clothes. And the potato — the machines were not designed for potatoes, the load of potatoes. So this is something where kind of people got the idea of listen to the voice of the customer, okay. They were just sending things back saying, "Well, your washing machine broke down after one year, one season of potatoes," right? They didn't say it was potatoes. You actually had to go out and talk to people about it.

I met a guy from Detroit, and he said that there was this one Japanese person — he was not a salesperson, but he was from one of the Japanese automotive companies. And the first question in the conversation, after the pleasantries, he says, "What kind of car do you drive? What do you like about it?" Okay. And this was just his way of trying to find out what the customer wants, okay, rather than "Will you buy my product?" Okay. It was sort of, "What do you want?" And they're going to try to make sure you get what you want. That's not all that different than the story I told you about Matt and how I told the LFM student to buy donuts and go out and ask the people on the factory floor what do you need help with. Okay. And some of the things were way beyond her capability of solving, but some were not, okay. And if you just start chipping away at the little things, you'll find that other people will help with the bigger things, okay.

So there was, back in the early 1920s, Dr. Shewhart was a statistician at AT&T Bell Labs — it was AT&T, it was just Bell Telephone back then — but he came up with a, what's now, some people call the Shewhart chart, some people call it a control chart or a run chart. And if you've ever worked in manufacturing, these are all over the factory floors of companies that are trying to use statistical process control. And it's not a terribly complicated chart. You basically are measuring some metric, okay, some measure, and you know that it has a lower control limit — which should be this, well that should say LCL, lower control limit. This is upper control limit. Your specification, your standard, says you must be between the two dashed lines. Your average is going to be right here on the blue line, and you actually make a measurement. It might be one out of a hundred parts, it may be every part, it depends on what you're producing. And you plot it over time, and you see when you're going — if you have a trend that's going to the lower control limit, or you drop out one time, you see when you're out of control or when you are heading out of control.

And there are all kinds of programs nowadays that use Shewhart charts and all kinds of statistics. You don't have to worry about these numbers, I just pulled something up off the internet. And here's the chart here. They plot the Gaussian distribution, or what they think is a Gaussian distribution. Here's the range, and this is the average — follows a Gaussian here, and the other, the range, does not. I'm not even sure what — they're defining range there, but I'm sure in Montgomery's book you'll find some of these details, because people have taken the simple concept of a Shewhart chart and they have added all kinds of mathematics and statistics.

Now, how useful is it in production? Well, when the LFM program started, there were nineteen professors at MIT, and they gave us term chairs, so I was a Leaders for Manufacturing professor. They said you're going to interact with Motorola, okay, fine. I went to Motorola, I met with the CEO of the company briefly, he was an MIT grad — that's one of the reasons MIT had joined the program, because he thought that MIT could help them. And I was there, I was from out of town, more than fifty miles away, I was there to help, okay. So it was an interesting experience, because then — my prior life as a consultant, or working with industry, I would always go to one factory of a company. They called me in, and they had a problem. Now I got to visit multiple Motorola facilities, and I got to see the differences between the facilities. I had seen that in the two years I worked for Bethlehem Steel, I worked with different facilities, and you had different plant managers, different philosophies, and that attitude from the top permeated down through. Some of them were Theory X type of managers, some of them were Theory Y type of managers, and you could very quickly tell.

And what I thought was interesting, I would go through one Motorola facility and say, "There be a Shewhart chart, a control chart pasted right there on the production floor." And I'd ask this person at the one Motorola facility, "What's this?" And they say, "Oh, that's something management makes us plot." And I said, "Well, what do you do with these?" "They're nothing, okay, it's not worth anything." Okay, that was his attitude. And in fact the attitude was, in fact, the fact, okay, because he didn't know how to use it, he didn't understand it, he did it because he was told. I go to another Motorola facility, he said, "What's that?" And the guy would say, "Oh," and he'd start explaining it to me in detail. I said, "Is it useful?" He says, "Oh yes, I mean, I can see what — my body tells me how" — he's using it. So just saying you're going to do something is not the same as actually getting it done, okay. So there are various types of statistics used, and that comes under the umbrella of total quality management.

Now, statistics. There's a famous quote — that I felt research on this morning, I've been using it for years and people attribute it to Mark Twain. I even have a book — it says this was a Benjamin Disraeli quote attributed to Mark Twain. "There's lies, damned lies, and statistics." By the way, who was Benjamin Disraeli, anybody know? He was the Prime Minister of Britain under Queen Victoria. He was Jewish, and it was very unusual for a Jewish person to rise to that level in the 19th century Britain, but he was apparently a very effective Prime Minister. It turns out I did a little research — all you have to do is go to Wikipedia, right — and it turns out no one ever has any record of Disraeli having written this. But when Mark Twain first used it, he attributed it to Benjamin Disraeli. Now, he may have talked to Disraeli and Disraeli may have said it in words, so it's not clear. There's some confusion about Disraeli — whether Disraeli said it or Mark Twain. But here, Benjamin Disraeli said, "There are three kinds of lies: lies, damned lies, and statistics." And the punchline is, nine out of ten people say that was Mark Twain. And I actually had a book once, and someone said — quoted it — and I said, "You know who said that?" And they said Mark Twain. He said no — I went to the shelf and pulled off my book, it says often attributed Mark Twain but actually Benjamin Disraeli. But if you read Wikipedia, which is even more detailed, it's not clear, but Mark Twain originally attributed something else. Anyway, that's just the punchline after the punchline.

And then here's something else. Good old Dilbert has to deal with statistics. "I don't have any accurate numbers, so I just made up this one. Studies have shown that accurate numbers aren't any more useful than the ones you make up. How many studies showed that?" "Eighty-seven." Which you can ask whether the eighty-seven was made up or not. And I think I told you my stories of numbers that I didn't exactly make up, but I estimated, and later I found that the Bureau of Labor Statistics was using my number because no one else — I was the only one who knew how the number came about, okay. And then there was another situation, I can't remember where, I had a number, I estimated, and I found other people were using it. It helps to be quantitative, but it's nice to know where the numbers come from there.

Anybody have any questions on statistical process control? There's another term which is actually part of statistical process control, it's all through Montgomery's book, called Six Sigma, okay. And Six Sigma was invented by a guy, Bill Smith, at Motorola in 1986, okay. It's a method for process improvement by reducing process variation. Remember, Montgomery says quality is the inverse of variability. So there's the variability word again. Jack Welch adopted in 1995 for General Electric across the board. And it seeks to get to Six Sigma, which is three point four defects per million. And if people just say, well, it's just the Gaussian distribution, it's the tail of the Gaussian distribution — well, if you think very hard about that, or you look at the error function and try to see where does that three point four come from, it's not so clear, okay. It's not just the tail.

In fact, I'm sure Montgomery explains it, but if you have a Gaussian distribution of your Shewhart-type measurements, okay, on the control chart, and that's the center one, and you presume that you're within, in this case, plus-or-minus Six Sigma, and then you say, well, let's say that I have a short-term measurement of my Gaussian, but I can have a long-term variability where this whole Gaussian is shifted one and a half Sigma up or one and a half Sigma down. And so you can talk about a short-term Six Sigma — or Sigma, which is just the width of the Gaussian — or you can talk about a long-term where you actually have, look at a long-term variation. And by making a bunch of assumptions, you can come up with the 3.4 parts per million. It's not exactly a simple thing. And there could be short-term and long-term Six Sigmas.

And so if you look — and again this is just out of Wikipedia, this is a chart in Wikipedia — Sigma level one, two, three, four, five, six, seven Sigma with one and a half Sigma shift, that's the shift of the Gaussian in the long term versus the short term Gaussian. Defects per million opportunities. A half a — one Sigma is just — that's one over e or something, right? Point six nine, the thinker know, anyway. But that is basically just one over the log of the natural logarithms, one over e. And it's sixty-nine percent defects, percent yield is thirty-one percent, yeah. And then he has a short term and long term, if you're really interested read Montgomery's book, okay. If you get down here to Six Sigma, it works out to three point four parts per million, or ninety-nine point nine six percent good parts. Well, everyone would love to be there in most cases. And that's this short term.

CPK — that's another term that they throw around in industry. How many people have worked in industry and have heard of a CPK before, anybody? Yes, you have. And what is it, what do you know — what does the process — yeah, process capability. And so a process capability, if I have a nice Gaussian that's well within my control limits, so I got a narrow Gaussian well within the control limits, then that's a good CPK. That's a CPK of two, which means Six Sigma. A CPK of one is three sigma, and that's like seven percent defects. So people don't always talk about, well, I'm at such-and-such a sigma, unless they like the Six Sigma term. More often people use CPK, process capability, and they'll talk about it.

I remember when I became department head in 1995, I basically — the first faculty meeting, for whatever reason, I was talking about, we needed to broaden the materials department and do other things like worry about manufacturing rather than just materials science. I was sort of on the far end of materials science, engineering, on the engineering side, and all the powerful people were materials scientists. They were the wannabe scientists who couldn't hack it in a school of science, okay, but they were trying to be scientists in the school of engineering. And I said, well, how many people here know about what a CPK is? And because Professor Clark was not there at that meeting, no one raised their hand. If Joel Clark had been there, he would have known, okay, because he works with people over at Sloan. But this was the technology that the CEOs of the top companies were adopting and trying to use. And that's why they had the TQM challenge, because the universities were still about twenty or thirty years behind where industry was. You can argue whether it's terminology, but in fact the Japanese were using this stuff back in the 1970s. In the 1980s they were killing us in the marketplace with quality, and American industry was just starting to learn from the Japanese, and they were using statistical process control.

And there's a number of gurus of total quality management. W. Edwards Deming, who won the National Medal of Technology for what he taught the Japanese, but he actually learned it originally under the U.S. Navy and some of the SUBSAFE programs and things. He was a professor at one of the New York City colleges, I can't remember which one. But no one would listen to him in this country about quality.

And I actually — I'll tell you this. Quality, my lesson on quality, when I worked for Bethlehem Steel — I don't even remember what it was, but I had come up with some process change they could do in the mill that would improve the quality at essentially no cost, okay. And I proposed this to my boss, who was an MIT PhD, who he said, "Well, why do we want to do this?" And I — well, I said, "We make better quality product." Now you have to remember, this was 1975 or so. And I had to spend a couple of days talking to him and lobbying him on why we would want to make better product, even if it cost us nothing. Why should we ask for a change? Okay. Well, he finally — convinced him sort of sheepishly, he went up to his boss who was a PhD from Lehigh University, and I — he got me an audience with the second-level-up boss. And I went in saying, "Look, this will cost us nothing, we can make better quality." He says, "Well, we don't want to do that." I said, "Why don't you want to make better quality?" Says, "Well, won't help us sell more steel." That was the attitude. Everything was how many tons of steel you poured.

In fact, at Bethlehem Steel in the 1930s, in the middle of the Depression, the head guy at Bethlehem Steel was a guy named Eugene Grace, and six of the ten top-paid managers in the country in the 1930s worked for Bethlehem Steel. And that was in part because Eugene Grace had a deal where he and his top management got paid a bonus for the number of tons of steel poured or cast, not on the number of tons shipped or the profit. So they sort of had the wrong metric, okay. But this was 1975, you know, thirty-five years later, and that same attitude of "it's how much you sell or pour that's important, not how much money you make" — excuse me, but that's why some companies are now bankrupt, Bethlehem being one of them.

It's also why — after — it's not the reason, but it's one of the reasons — after I was there for thirteen months, I asked myself, and this is a question you should ask yourself about once a year or once every two years, where do you want to be in five years? And the one clear answer after thirteen months at Bethlehem Steel was, not at Bethlehem Steel. I mean, I'd rather be anywhere than Bethlehem Steel. And so I started looking for a job, I called up some old professors, and a couple of them said, "If you ever thought of being a faculty member?" I said, "Oh no, I wouldn't want to take that job, no, who wants to be an assistant professor and get dumped on?" Okay, anyway, that's another story. But I came back here, they convinced me.

So I also say I often tell students that I've had to reinvent myself every three years, okay. If you ask yourself where you want to be five years from now, that is a moving target through your career, and you have to stop and say — I mean, I had a reputation as the guy who could do anything in the laboratory as an undergraduate. What that meant is I could clean the ink out of the jet printers — they weren't inkjet printers back then, but they were little tubes that distribute the ink on the strip chart recorder, and the graduate students never cleaned them, and they were only two of these strip chart recorders in the lab with like twelve graduate students. And they'd always send me to go get one, and they were always clogged. All you had to is take a little wire, cleaning out the dried ink, it worked, and everybody thought I was wonderful because I could clean up any tube. Whoo.

Anyway, but when I became an assistant professor, I knew the last thing I wanted to do was spend my time cleaning ink out of ink pens, okay. I had to stay at my desk, I had to write proposals, you know, it was a different job, a different set of responsibilities that I had. And so I could explain to my graduate students or the undergraduates how to clean the pens, okay, whatever, that wasn't the problem. But I've had to advise assistant professors for years on what they should be doing and what they shouldn't be doing. And what you were doing as a graduate student that you were very good at is not what you should be doing when you take your first job, necessarily, okay. You have to kind of say what carries over and what doesn't.

So one of the things that I learned after five years working with Motorola and other companies was, if you were at three Sigma, you couldn't apply the statistical process control because you were just too far out of spec. I mean, if I go back to the chart, okay, three Sigma, I'm producing seven percent defects, and there are multiple reasons. When you've got that many problems, there are multiple reasons. And so trying to apply statistics, which tends to home in on one, is just going to give you lousy statistics, because you've got too many variables. So you've got to fix your process if you're at three Sigma or below. Between three Sigma and five Sigma, SPC works no problem.

But I learned that many companies would use SPC and they'd get to five Sigma and they'd have this goal to get to Six Sigma and they really couldn't do it. And the reason was, at Six Sigma there's not enough defects to give me good statistics, okay. In fact, now we're talking about something that's not a systemic defect, it's sort of some isolated thing, you know, someone dropped their coffee into the machine or something, okay. And you produce, you know, ten bad parts or something, and you get some problem. So what you have to do is when you're at Six Sigma and above — there's too little data — in fact, for Sigma, maybe, except — you have to also recognize that Six Sigma is just an idealized goal.

I always said, if you're trying to make bricks for a barn or a house or something, four Sigma may be, okay. But this was — I started saying this back in the '90s when they had the problem with the Hubble telescope. Anybody know what happened with the Hubble telescope? Two billion dollar device to look at stars and probe back into the earliest ages of the universe, and they had a problem when they ground the mirror of this two billion dollar machine, that they screwed up, okay. And it wasn't as nice and parabolic as it should be. Well, the problem is, if you're making Hubble telescopes, it's not easy to go fix it when it's up there, okay. So you really needed process controls that were much better than even Six Sigma.

Now it turns out they got out of the problem. Anybody know how they got out of the Hubble telescope problem and got it back working? Yes — in a sense, but mostly it was — you're right in principle, but it turns out, MIT Lincoln Lab — as part of this story — had been working for years to improve the surveillance of, you know, satellites and aircraft taking pictures of all those things in Cuba and around the world and seeing things. And they were always trying to get better and better resolution. And they basically had come up — they would shoot a laser beam through the atmosphere and measure the distortion of the laser beam. They would then back-calculate — this is the inverse problem, which is a very difficult problem — how to correct the light, you know, the distortion. So they knew the — later — oh, Hubble telescope had a distortion problem. They could measure how it was wavy and where the distortion was, and they just applied the physics that Lincoln Lab had developed for their super-secret, you know, surveillance satellite and federal imagery and stuff. They applied it, and now you could get a crystal-clear picture from the Hubble telescope, like it was supposed to produce, even though it was defective.

So there are oftentimes a work — a way to work around the problem. That's the story of what was it, Apollo 13, that they had to come back and make up their — a little oxygen or CO2 thing. I mean, these guys are gonna die. Who was — Tom Hanks, why did — he was about to die, right? You know, I think it was Apollo 13 was the movie. They talked about the true story and how the NASA engineers had to use duct tape and other things. They had to figure out what was available on that lunar module or whatever the thing — they were going to re-enter — and how they could make something that would regenerate the oxygen out of — regenerate the oxygen out of the air. So there's often another way to work around it. In this case they worked around that.

But at the time they did that, there was plans for other telescopes beyond the Hubble. But it turns out they no longer need them. They apply that same technology in the opposite direction, and now land-based telescopes can see as if there was no distortion from that material is — again, they shoot a laser beam through the atmosphere, measure the distortion in the atmosphere, correct for it mathematically, receive the light and do the correction, and you get crystal-clear images.

Yeah. [Student question, inaudible — appears to reference Connecticut.] It could be, I mean, I hadn't heard that story, but that sounds like the type of thing — how do you deal with how many people you're going to let into a super-secret problem? Well, that's a policy decision that's above my pay grade, okay. I don't know that I have an answer for that. You can't put on insecurity — you can't put all the information in one person, because if that person becomes insecure — then are not insecure but unsecured — you got a bigger problem. So what they've done is they parse it up so no one knows all the pieces of the puzzle, everyone's got one piece of the puzzle. But eventually someone has to integrate the puzzle, okay. So now we have people in the White House who can do that but none of them can get a security clearance. It's — anyway, I don't have an answer for that.

Although I remember a story of one of my first consulting cases. And the CEO — I wasn't there, but they were fighting about something, and the CEO brought out a little turtle that had a little kind of bobblehead neck on the turtle, and he put it on the table and he said, "Sometimes you have to stick your neck out, you have to take a risk," okay. And you just have to decide what type of risk you're willing to take. Any other questions on that?

So Six Sigma. If you are at Six Sigma or above, if you want to go and get better, you have to design a more robust process. And that's the whole engineering design process of defining a problem, identifying things, brainstorming. And we'll talk about some of the tools that people use to do these — the whole things in engineering design that became very popular in the 1990s, and some term called design for manufacturability. If you go to Wikipedia and look at their total quality control website, they will have a whole series of quality websites, one of which is design for manufacturability. And there were two professors down at University of Rhode Island who basically kind of — something, then — can't remember the names right now. Anyway, these two guys came up with the idea you should design something so it's easy to assemble, okay.

There were always problems like, was that the Chevy Vega back in the 1970s, that they had people designing the engine, they had people designing the structure that it went into, and it turns out they assembled the engine and then they would just put the whole engine into the vehicle on the assembly line. But when you needed to change the spark plugs, you had to remove the engine to change the spark plugs, because someone had a little interference like the structure of the vehicle. And so there's big problems with that. There were people at MIT that I know, some of whom were retired now, but Chris Magee — Chris Magee [Chryssostomidis] — and Nick Patrikalakis, who are now in mechanical engineering — Chris is retired, but that was what they worked on. They were working on developing the CAD-CAM processes that would make sure you don't design two things in the same space, okay. And so you'd have ways to control that. So there is this whole field of design for manufacturability. Reduce the number of parts, supposedly, if you can.

Yes? [Student question.] In designing a submarine or some ship, you have pipes for water and you have pipes for oil, for example, and they may have had — both of them going through. And once Chris crossing the other, you can't, you know, nature doesn't work that way, okay. But it's actually a common problem. And if you get into — I had a student from the — what's now LGO — working at one of the submarine shipyards in Britain, can't remember, was the Vickers or — anyway, somebody — there was somebody that had joined the LGO, and they owned one of the shipyards, and she went over. And it turns out she was put in the piping shop for specifically that problem. There were, you know, people would look at the drawing, and they'd find, well, this is impossible to build, what's on the drawing doesn't fit, okay. And they had various ways to measure things in the field. But then you had this piping had to use this metal and that piping had to use that metal, and they're always out of the — of inventory, because when they had a problem, someone would steal from the inventory that already been ordered, which might be a nine-month lead time. So they get that problem solved, but all they did was create other problems for the future. And as a result, it took them ten years to build a submarine when it should have only taken three, okay. That's a problem.

And we ran into it — I ran into it at Motorola, you know, they have all these different components that go on a circuit board, and someone orders the right number of components. Well, how much extra stock do you need in case one of those components is bad? Because if you order just the right amount, you know, and one of them is bad, then now you can't finish all your circuit boards. And this was actually the biggest problem in making circuit boards that Motorola was finding — the inventory to build it — even though you had all kinds of people who were there trying to make sure they had all the stock ordered with the right lead times and everything else. So there, it's a very complex process. And this is part of what they now call supply — a supply chain management, okay. But that gets into other things.

Now, any other questions? Did I tell you the story about the Ford Mustangs and pushing them out? I did, I know, okay, I did to the class, okay, fine, I won't tell you that. I actually had a slide of that. So this is the Ford Mustang plant where if the cars didn't start they had to push them out. The manager said you got to push them rather than bring in a tow motor. And this is the Boynton Beach where the manager says no more rework facility, you're going to fix the problem rather than make five percent bad parts.

So now I'll talk about The Goal, or the Theory of Constraints. And this is one of these things that creates the title of in — TQM, in my opinion, okay. Eli Goldratt was a officer in the Israeli Air Force or something, and he wrote this book called The Goal, okay. And has anyone read The Goal? Yeah, what do you think of it? Yeah, I know — that's why I had to read it though, twenty-five years ago. Yeah, there's better ways to spend that much time, how about that, we agree on that, okay.

So anyway, it's a little story. There's this guy who's trying to worry about his manufacturing plant, and he's having all kinds of problems, and he takes a bunch of Scouts on a scouting trip. And there's another guy who's a consultant, I think he meets Jonah in the airport or something, and Jonah says, well, you have to follow the goal. And all through the both the book, Jonah is telling him, as his advisor, you have to think of the goal. And do you remember the goal of the company is to make money? See, it's just a memorable book, right, I mean that's the great thing about — I remember Herbie, right.

So Herbie was the fat Scout in this metaphor. It's a long metaphor, okay. In fact, this is how long it is, but it's that he sold like two and a half million copies or something of this thing. So he's dead now, but he's made a fortune off of it. And the idea is, well, Time magazine says it's one of the five most influential business books. Well, okay. And he comes up with the Theory of Constraints. And the problem here was Herbie was always at the rear of the line, and they were trying to march through, and they always had to go and help Herbie. Okay. Eventually Jonah gives him some advice, and he realizes if you put Herbie as the leader in the front, then everyone else will follow behind Herbie and you'll go at the right pace, which is Herbie's pace, okay. And so you follow the Theory of Constraints — Herbie was the constraint.

So I have a story about Herbie and Ford, okay. The Theory of Constraints. This is the air conditioner for the F-150 pickup, and I had a student at LFM, LGO, whatever it is, who was assigned to this plant where they had one line — that had actually about twenty lines, but each line made one type of air conditioner for one Ford product, and this was the F-150, which is the largest-selling vehicle in the world, okay. And they were selling like hotcakes in the '90s, they still sell pretty well, and they were reaching their capacity constraint on the air conditioner assembly line for the F-150. And it was going to cost twenty million dollars to build a new line, and then those would both be working at like fifty-two percent of capacity, and so that wasn't gonna be very efficient. How could you get more throughput through? Ford was a great believer in Theory of Constraints at the time, this is mid to late '90s, I don't remember exactly, I think it was the later half of '90s.

And so we went out there and they showed us this problem where the student was going to spend seven months. And if you went in the plant manager's office, every table was empty except one by the couch on the coffee table — there was a copy of The Goal. Only thing anywhere. I mean, this guy's managing, you know, a fifty million dollar a year plant or something, and he's got one thing — there's The Goal. He lives by The Goal. This is one of the big problems of American management — they get on a kick and they think it will solve all problems. They think you do not have to think your way through a problem, you can manage your way through a problem with some heuristic. It doesn't work, okay.

So it turns out they had spent a quarter million dollars on consultants who were experts in Theory of Constraints coming in looking at this thing, and they decided they would throw this problem to this MIT student for seven months. And it turns out what they had told these — these consultants had told Ford was to introduce a constraint into the system to get more throughput through. Excuse me — does that make sense to anybody here? You're gonna constrict the pipe in order to flow more liquid through it, okay? That's what they're saying, 'cause Theory of Constraints teaches you this.

So — what do I — I told you my story about how you figure out how to solve a problem, you just find the time to go out there and ask the foreman, right, because he knows what the problem is. I go out and ask the foreman, and I say, "Where's your constraint?" He says, "It's where I put the people." I said, "What do you mean by that?" He says, "Well, I got people who work very quickly, and I got" — because there are twenty people actually doing sort of hand assembly, putting together, I mean, not really — you still had to put things together, you didn't have robots everywhere, they had some robots, but they had twenty people on this assembly line. He says, "We shift them around every day so they don't get just totally bored doing the same thing five days a week. And some people work faster than others. And so my constraint is the people, not the machines." Well, that made sense, okay.

So anyway, that gave the student something to think about and to frame their thesis around, to try to improve the throughput. The goal in that case was not to make money — it was make money for Ford, but just making air conditioner compressors was not going to make money unless you got all the vehicles out there to sell. It was a subset of the overall problem. And this — all these high-priced consultants were basically a drain on the system, okay.

Now, Eli Goldratt became very popular. He came here in the early '90s, and I was invited to listen to his lecture and to go have lunch with him — a bunch — with a bunch of the other LFM professors. And over the — they have a restaurant — he used to be a faculty club on the top floor of the Sloan building, a sixth floor, and we sat in this conference room. And so somebody asks Eli what he was working on now. He was working on figuring out how to manage governments, like the U.S. federal government, and he was going to find the constraint and he was going to manage that. Now, the federal government is a trillion dollar operation, is there only one constraint in the federal government? That's his assumption, right? He was gonna find that constraint and he's gonna — well, that makes life easy, right? So Eli Goldratt was a paragon of simplicity and irrelevance, okay.

So that's — it's not as if I — I don't think he's a wonderful person, made a lot of money, but anyway. And we'll talk tomorrow about the awards you can get by being a certified Jonah, or a Six Sigma Black Belt, and other things, which are other buzzwords, and how there's a whole group of consultants who have grown up around these buzzwords. Which is why some people think — okay, thanks, see you tomorrow.