WIE_F2015_10

How to be a Successful Engineer Fall 2015 Session · 9 sections 12 cases · Watch on YouTube ↗ all files
Layer 3 — readable edition

§1. Edward Tufte and the display of information [00:00]

§1.p1

In the back there is a handout from Edward Tufte. It's chapter two of one of his books. He has four books. Remember he's a professor at Yale, but he started getting interested in how to display information. For example, this he considers the best graph ever made. If you take his one-day seminar, I think it's five hundred dollars, they'll be very — oh yeah you have to understand it — but I won't spend a second with it. He basically looks at lots of different ways to graph things, and sometimes you can graph things and actually have no value whatsoever to your graph or your table, and sometimes you can have all kinds of information and put it in one thing.

§1.p2

Tufte has gotten into giving these seminars for industry. I took it over at the Copley Plaza Hotel once. You pay five hundred dollars for one day, there's 200 people, so he's collecting a hundred thousand dollars for the day. Not a bad salary, although he does have to pay for the room, which probably cost him ten thousand. He's made so much money off selling these books and giving these seminars that he has Sir Francis Bacon's copy of Newton's Principia. You get to see the pages, because he has underlings in white gloves carrying the book around showing you the page he wants you to see. He has a lot of these rare books because he likes to collect rare books. He has a lot of good ideas. He's also one of the most arrogant people I've ever met, but aside from that.

§1.p3

This is Napoleon's march from Paris to Moscow. You can't read the stuff up here, but it's in French anyway. Here's Moscow in French, and this should be Paris over here. The width of this is the size of his army going out. Then little pieces of the army went off to the side and got sick and went home. They finally get to Moscow, and here's the army coming back, and you can see the size of the army compared to what it was when it left. So there's a lot of information on this graph. And then there's the temperature plotted down here on the days that they were marching out. It was a cold winter. He considers that one of the most useful graphs — I can't remember exactly, it's been fifteen years since I took his course.

§1.p4

He's written four books. He publishes the book himself. They're all on 100% vellum acid-free paper because he considers them to be a work of art that you will have forever. But they're very inexpensive too. The books are like sixty bucks apiece. However, this little handout cost five dollars. It's on 100% rag paper, acid-free, so you will have it forever. It will not deteriorate. I went up to him at the lunch break and I said, Professor Tufte, I'm a professor and I'd like to use this — if I buy in bulk, like five hundred, could I get a discount? Well, I guess you'll just Xerox it anyway, so I guess I'd better do it. So I got it for four bucks. It was still 1,600 or two thousand dollars that I paid. I've been using them for a few years for various classes.


§2. PowerPoint and the Gettysburg Address [04:14]

§2.p1

So we've been talking about decision making. Engineers have to make all kinds of decisions, trade-offs about different things. They also have to learn to communicate what they say, and so communication is supposed to be part of what we're doing here. So here it is: visual and statistical thinking, displays of evidence for making decisions. He has two examples in here. This is chapter two of one of the four books. The first book is the visual display of numbers. Another one he calls the visual display of words. Another, the visual display of verbs. I don't remember what the fourth is. But he has a nice succinct way. He's a good teacher.

§2.p2

And here's one of my favorites: the PowerPoint version, in auto-content PowerPoint, of the Gettysburg Address. Have I shown you this before? This is the Gettysburg Address reduced to six PowerPoints. The organizational overview was eighty-seven years ago versus now. You graph that. Agenda: met on a battlefield, it's a great battlefield, dedicate portion of field, it's fitting unfinished work, great. A review of key objectives and critical success factors: what makes nation unique, conceived in liberty, men are equal, shared vision, new birth of freedom, government of, by, the people. Not on the agenda: his dedication, consecrating, hallowing in a narrow sense to add or detract, nor remember what we say. Summary: a new nation, civil war, dedicate field. He took a great piece of English literature and PowerPoint reduced it to pure drivel.

§2.p3

I use this as — now that you're getting your PowerPoints ready for your presentations — PowerPoint is a great crutch for people to present their outline, or to read from it. The best lecturers, the ones who make lots of money like Lester Thurow making thirty thousand dollars a lecture in 1988, or Tufte making a hundred thousand dollars for a day's teaching — they don't use PowerPoints. I've talked to you about that before. You really want the audience to focus on you and what you're saying, rather than up there looking and trying to figure out what the graph means. In fact even Tufte, who thinks the graph of Napoleon's march to Moscow and back is one of the greatest graphs, he wouldn't expect it to be a useful thing to just put up. Because just like Catherine said, what's this? It's too complex. But once you actually get a little bit of an explanation — oh gee, that's pretty neat. All of a sudden you have a visual image of how many people got lost and where. And then he shows you some of those lines where they were crossing a river that was frozen, and people fell in and froze.


§3. The scientific method and confirmation bias [07:54]

§3.p1

But I want to take you through the scientific method. We talked about the scientific method before. Do people remember the four key points of the scientific method? I wasn't that impressed with how many people rattled them off. You first collect the data. The reference I gave you started with defining the problem, but assuming you know what your problem is, you've got to collect data about it. You have to analyze it. Develop a hypothesis and test it. And only then can you draw conclusions.

§3.p2

Now this has been around for thousands of years. Aristotle and others used it. If you develop a hypothesis and test it and it doesn't work, you've got to go back and develop a new hypothesis, and you keep on doing this until you develop a hypothesis that cannot be refuted. You're trying to falsify your hypothesis. Your test should be designed to show that your hypothesis is false. If you do your testing and you cannot show it's false, then the conclusion is it must be true. There's a quote from somebody famous — I can't remember who — when we've tested something, the only thing that remains, no matter how improbable, must be the truth, if we've eliminated all the other possibilities. Maybe it's Sir Arthur Conan Doyle — good old Sherlock Holmes. Used deduction to solve the problems.

§3.p3

Now there's also induction. The thing I gave you, about three or four pages on the scientific method, is the most complete description I've ever seen. It talks about confirmation bias, where people draw their conclusions in the beginning too early before they've laid out enough hypotheses to test, and they don't eliminate all the other possibilities. For example, the thing out of Kahneman's book, where someone had looked at the data and they found kidney cancer was worst in the poor rural counties. You can define that as the cause. It also turns out, if you look at the data another way, you find that kidney cancer was least in all those counties. So it was the worst and the least, depending on what county you're talking about. That was, as Kahneman said, the statistics of small numbers — if you're not averaging over a large enough cohort.

§3.p4

There's expectation bias. You think that is the cause, and so you end up proving it's the cause. Confirmation bias I often call the reverse scientific method. You draw your conclusion, you throw out all data that doesn't fit, and you go backwards. It's circular reasoning. You start with your conclusion and end up concluding the same thing. You're just going around in circles. You don't accept any data that doesn't fit. So if you avoid confirmation bias, you will test all the potential theories.


§4. John Snow and the Broad Street pump [11:40]

§4.p1

So Tufte gives us two examples. One is John Snow. John Snow was actually the anesthetist to Queen Victoria. Why she needed her own anesthetist — have any of you ever been anesthetized? Anyway, he was a medical doctor. Tufte likes to point out that depending on how you display your data and analyze it, you may get better results than others. He gives us a successful story and an unsuccessful story.

§4.p2

What happened was cholera broke out in the Broad Street area of central London in 1854. John Snow had investigated earlier epidemics and suspected the water from a community water pump was contaminated. This is before we really had clean water. In the nineteenth century people were lucky to have clean water. People didn't understand bacteria in 1854. Tufte talks about what some of the theories were — that you contracted cholera through the air, things like that. But Snow had a hypothesis. He had a good idea, and he was going to use that hypothesis to test the data, and then he was going to evaluate things. He goes through the whole process of considering other things even after you think you've proven things.

§4.p3

At the bottom is a graph showing the death rate starting on August 31st. People started dying, peaked the next day, and then it started to wane. Here's the cumulative death rate. Six hundred people died during the same epidemic of only a few days — about a week's time. You've got some of these graphs if you picked up one of these things in the back. What he did is he plotted on a map of London, and it turns out right here in the centre there's a little dot. It says "pump." This is the most deaths. Each one of these represents a little line, and the longer the bar — you can see the individual lines. This thing's not sharp enough to distinguish them. You can think of each one as a little coffin. He has a picture later of little coffins.

§4.p4

It turns out on September 7th, Snow talked to the local town officials in that part of London, and he explained that he thought it was the water that everybody was drinking from the community pump. The next day the town officials took the handle off the pump, and the death rate decreased. Now you have to consider alternative explanations and contrary cases. For example, it was already decreasing before he even did this, right? So you have to say, well, why is that? Maybe it's not the right hypothesis, if it already had started to decrease. But in fact what had happened: everyone was fleeing the neighborhood. After two days of a hundred and eighty people dying in one day, or whatever the number was, a lot of people decided to go visit their relatives in the country. Good idea. So you didn't have a constant population, and if you could have corrected for that population, you may have found that it was staying constant.

§4.p5

He gives a lot of anecdotal examples. There was this one elderly woman who loved the taste of the water from the Broad Street pump. Once a week her nephew or son would go into London and bring back some water from the Broad Street pump. He did, and she died the next day, even though she was a long way away from central London. Cholera may take two days, but if you're elderly it may take only one day as you get sick. The stories in here, I don't remember all the details, you can read it — it's not something in my general knowledge. But he goes through a couple of pages to talk about how they tracked down some of the outliers. As you can explain the outliers — the ones that seem to disprove your hypothesis — you find that those have an explanation too. All you're doing is reinforcing your hypothesis.

§4.p6

He points out that you can plot the deaths on a daily basis, the same plot we showed you before, or you can plot it on a weekly basis. Sometimes when you aggregate things — in this case you still see the decrease — but sometimes when you start averaging things you can get really bad data. He gives you another example from something else that shows you've got a little peak here, but if you average over a certain amount, you lose it. Averaging is something that electrical engineers and mechanical engineers do a lot in signal analysis. They may have a moving window, or a Kalman filter — different types of ways to filter the data to take the noise out. Sometimes you can take the real data out too, if you filter out enough; you just get a straight line. And here's the little plot of Queen Victoria, John Snow, and a plot of coffins. You can read the whole story yourself. It's actually good reading when you're doing nothing.


§5. The Space Shuttle Challenger and the space-laser aside [17:56]

§5.p1

The next one is the Space Shuttle Challenger disaster, 1986. Here's the space shuttle. It has two solid rocket motors, one liquid oxygen / liquid hydrogen tank — the center tank, which is expendable. The solid rocket motors are actually recoverable. They have a parachute, they eject them and they fall into the ocean, and they recover and reuse them. The main tank was not recoverable. It went further out into space and got burned up in the atmosphere. But the solid rocket motors were designed to be reusable.

§5.p2

In order to be reusable, these things are so big you had to build them in modules, and you had to have o-ring seals — well, you didn't have to, you could have done something else, but they had a double o-ring seal. There are different pictures of this in the booklet. Good old NASA has things down to fifty-eight point seven-eight seconds. Wow, good, I'm glad they know it's that accurate to the millisecond. You can see a little white spot at one of these o-ring seals as the solid fuel was coming out.

§5.p3

Student: [Disadvantage of solid rocket motors?]

§5.p4

Once you ignite it, there's no way to stop it. You can't put it out, because it's a solid fuel that contains its own oxidant and fuel mixed together. It doesn't need air. Liquid oxygen, liquid hydrogen — you've got valves, you've got a flowing liquid, you can shut the valve and all of a sudden you can throttle back your energy. So the solid rocket motors — that's pretty blunt-force rocketry. You light it and you hope for the best. Whereas something with a liquid, that you can meter, has certain advantages. Obviously as you're getting up there and you're starting to maneuver, you'd like to change your power, and that's why they have a liquid-fueled rocket.

§5.p5

They knew what had happened was blow-by the seals. NASA basically shut down — they do that after every launch. NASA basically shut down for about two and a half years. There were no shuttle flights for two and a half years. What is the main purpose of the shuttle? What do they tell the public, and what do they tell the hidden secret sessions of Congress?

§5.p6

It's a shuttle, just like the shuttle from here to New York. That's why they called it the shuttle. At the time it cost ten thousand dollars a pound to put a payload into orbit. It still does. The shuttle was advertised in the early 70s as going to drop the price by a factor of ten — it would only cost a thousand dollars a pound to get into orbit. People said, oh, we could colonize the moon. Yeah, twenty thousand dollars a pound. It would pay to have someone go on a diet before they go on their flight.

§5.p7

Student: They had schedules.

§5.p8

And as you said, the real justification was military. They had to get payloads into orbit, they wanted to be able to get there and work on them — like the International Space Station. But they had military objects that were hush-hush, like space-based lasers. Just by coincidence, Rockwell International had the contract for the space laser weapon. There are various ways — there's things that are classified. Unfortunately before I ever had a security clearance, I didn't have to worry about what I figured out, because I never learned it through classified means — I didn't have a clearance.

§5.p9

I had figured out through three different things that the power of a space-based laser weapon was five megawatts. The Air Force was trying to get a 50 megawatt superconducting generator, and I'd worked on superconductivity in the early 70s. The lasers — the chemical lasers — were about 10 percent efficient. I can multiply 50 by 0.1 and come up with five megawatts as the laser power. Professor Bowen was working on one-meter-diameter laser windows for Lincoln Laboratory, forging these things out of potassium chloride. No one had ever tried to make very high purity, low imperfection — no dislocations or voids — things of potassium fluoride and potassium chloride and cesium chloride. No one had ever tried to do that before. That's because the wavelength of the laser — this was transparent to that particular wavelength. You could look in the literature and find the damage threshold per square centimeter of laser power that you could go through it. You could multiply it by one meter, which is what he was trying to do — separate contract — put the two together and come up with five megawatts. And then there was something else, I don't remember what the other one was.

§5.p10

So I asked a friend of mine who worked in the aerospace industry in California. I said, Charlie, do you know what the space-based laser's power would be? I figured it out and I told him my two methods. He says, well, we have the main contract for that, so I really shouldn't tell you, but let me tell you that it's a pretty good estimate you've got. And what's more, it fits perfectly in the cargo bay of the shuttle. This was 1978 or so.

§5.p11

There are other things. I was talking to a woman who lives around the corner — her son is working at Lawrence Livermore on supercomputers. She went to visit Lawrence Livermore Lab. As you know, Lawrence Livermore Lab is one of our weapons labs. I visited there once, because I had a student doing a thesis — he's still at Lawrence Livermore. Every building I went in had one of these little triangular temporary signs that would say "uncleared visitor today," which meant everybody had to keep their file cabinets locked while I was there, even though I had like three people escorting me everywhere I went. The only time I was less safe than that was when the KGB was following me in Moscow and Kiev. And then I was safe as long as they didn't turn on me. I had a police escort wherever I went. Well, at Lawrence Livermore I had a security escort wherever I went.

§5.p12

Anyway, this woman around the street is a virtuoso violinist. She's not technically inclined — her son is the computer scientist. She started telling me about this big laser. I said, you mean NIF? Oh yeah, that's it. They're doing fusion. She's pretty bright; she asked the question: how much power have you gotten out of this? They tell her, well, nothing yet, we haven't reached breakeven. You've spent all these billion dollars, have been working for fifty years? Then I explained to her, the real reason for the National Ignition Facility is to simulate nuclear weapons, because we can't do actual weapons tests anymore. So they built this five-billion-dollar laser, or set of lasers, so they can hit something small and implode it, get nuclear fusion, and test the physics of the weapons. So a lot of things are dual-use. They tell the public one thing — we're trying to make free energy by nuclear fusion, which they sort of are — but the main reason they're doing it is to simulate nuclear weapons effects.


§6. Morton Thiokol, the o-rings, and the politics of the launch [27:40]

§6.p1

One day before the flight the predicted temperature for the launch was 26 to 29 degrees Fahrenheit. The engineers who designed the rocket opposed the launching.

§6.p2

Student: Tell me more about that part of it.

§6.p3

They're in here. They sort of presented the wrong data. They pushed the data on blow-by — twice before they had some soot get past the o-rings. It's 4,000 degrees on one side of the o-ring and whatever the ambient temperature is up there — minus 30 degrees in the atmosphere — on the other side of the joint. Here's a close-up of the joint. You've got the book, so I'm not going to blow all these up. Here's the unstressed joint. When things get hot, that's why they have to have two o-rings, and you can get blow-by one or two.

§6.p4

Tufte, who likes to be critical, said one of the problems with these presentations — here are the handwritten presentations that Morton Thiokol, the designer of the solid rocket motor, who was a contractor to NASA — they didn't provide the names of the people who prepared the material, so no one was going to be responsible for this. The engineers at Morton Thiokol, which was out in Ogden, Utah, had been successful in 24 flights. The engineers went to the Thiokol managers and said, we don't think you should fly tomorrow. They'd already postponed it once. Do you know any of the politics about the flight? Thiokol was in a position where they had a 1.3 billion dollar continuation contract for the next set of shuttles. So the upper managers really wanted to please NASA.

§6.p5

On the other part of the political spectrum, this was the flight that Christa McAuliffe [Krista McAuliffe], the teacher from New Hampshire, and a few other people were on. It just so happened Ronald Reagan was going to give his State of the Union that night. They were talking about, potentially once they're up there in orbit, he's talking to Christa McAuliffe during the State of the Union address. Everybody says, we didn't postpone it for that, we would have postponed it even if we would have missed that opportunity — yeah, in some people's minds. But in any case, it was still there.

§6.p6

There was this engineer named Boisjoly [Bosco Lee] who was one of the ones who helped prepare some of these slides and said, you can't do this, we have too high a risk of failure. The Thiokol engineers basically take that and they water it down. They call NASA in Huntsville and say we don't recommend launching. NASA called back, or faxed back — this is days of faxes — and said, well, you've never told us not to launch before. What does that have to do with any of this? Sort of a sunk-cost problem. There was all this politics going on. There have been whole books written about this. [Boisjoly] — they actually had a conference call scheduled. They were in the conference call, and it wasn't going very well for the decision not to launch. NASA was trying to push Thiokol, and the Thiokol managers were trying to push their engineers into saying everything was okay. So they finally decided to cut off the conference call and just have the Thiokol managers talk to the engineers. One vice president of Thiokol told [Boisjoly] — I think his name was Bob — hey, go take off your engineer hat and put on your manager's hat. To me, that's — I told you engineering and management are basically the same. You can't take off one hat and put the other on. But they pressured [Boisjoly]. Everybody's telling [Boisjoly], oh don't worry about it Bob, it's no problem, everything's fine.

§6.p7

They plotted it wrong. They plotted blow-by and they didn't plot — they actually did have the data on there. This is one of the plots, and they have the temperature of the expected day. This is SRM 25, the 25th shuttle flight. They had 24 successful shuttle launches, and NASA had done a multimillion-dollar study showing that the probability of having a failure on liftoff of a shuttle was one in 10,000.

§6.p8

In the investigation later, Richard Feynman, the Nobel laureate from Caltech — he'd started out at MIT, his career went downhill from there, he went to Princeton and then Caltech. He graduated from MIT with a bachelor's degree in physics. But MIT Physics Department will not admit their own students. They think there's plenty of other good physics departments — and that's true, there are plenty of other good physics departments. So he went to Princeton, and he wrote books about the difference between Princeton and MIT. Then he went to Caltech, and he came up with quantum chromodynamics [quantum electrodynamics] and won the Nobel Prize. He was sort of a spokesman for science. At the hearing, when NASA was saying, this is a very rare event and we estimated the probability is 1 in 10,000 for a failure, Feynman said, well, I went back and looked at all the shots since World War 2, when we started rocketry, and we've had failures on 4 percent of our rocket flights. This was the 25th shuttle flight, which is sort of coincidence, because anyone knows if it really was 4%, but we won't get into the statistics of that.

§6.p9

They had temperature of the o-ring — this is history of temperatures — but the slide before was blow-by history, where you got a little soot past it. The o-ring history was a little different. They actually had o-rings where one of the o-rings basically burned up, and so you only had one o-ring left as the seal on some of these. Two previous launches had blow-by, and then they had some o-ring damage on SRM 15. SRM 15 was the next-lowest-temperature flight they had ever had: 53 degrees versus 29.

§6.p10

On the next page Tufte gives two ways. You could do it in tabular form. If you know that temperature is your important variable, the coldest launch they'd ever had was 53 degrees, and they had a damage index of 11 on the o-rings, and they had only had two blow-by incidences, one at 53 and one at 75. You can look here and you see there's sort of a cluster of o-ring damage, and then there's a few that had some o-ring damage — they lost the rocket motor case in the ocean on that one. You can plot it as a graph. Here's the index of damage, going up like this. You've got some data points here, but if you take all those, you kind of see this curve going up. Here's our temperature over here. If anyone had looked at this graph the day before the flight, they might have said, hmm, maybe we should look a little closer at o-ring damage. It becomes obvious in hindsight, which is part of what all this is. But what he's pointing out is you really do need to look at the data the right way, or you can miss the important parts of it.

§6.p11

Feynman was testifying before Congress, and he went out and he asked for a cup of ice water. They thought it was because he wanted to drink it, but in fact he had gone to the hardware store and picked up a C-clamp, and he had a piece of rubber tubing. He clamped it, put it in the ice water, got it cold, took it out, undid the clamp — you could see it didn't spring back to its original shape right away. There's no resiliency at lower temperatures for rubber. That's how he demonstrated it. Tufte talks about how even Feynman knew that was sort of a Mickey Mouse experiment, but it still showed the point.

§6.p12

It talks about the difficulty for the rocket maker to deny the demands of its major client. NASA was really pushing for the go-ahead. There was the possibility of a televised conversation of Christa McAuliffe with President Reagan. Feynman said in the final report: "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Some of the stuff in the book by Kahneman, Thinking Fast and Slow — you know, you have your intuitive thing. Thiokol intuitively noted, we never tested these things at 29 degrees Fahrenheit. In fact, we thought we were going to be in Florida — Cape Kennedy, Cape Canaveral. In fact they usually were, but it was an unusually cold time at Cape Canaveral. I'm sure we probably lost a billion dollars worth of oranges that week.


§7. The Cambridge courthouse aside and the Columbia [38:40]

§7.p1

So there's a successful case with John Snow, and an unsuccessful one with the space shuttle. The other thing about the space shuttle: engineers — [Boisjoly] resigned from Morton Thiokol, and he went on the lecture circuit. He used to say he wished he had stuck to his guns and told the managers that they were wrong. But he took off his engineer's hat and put on the manager's hat, and it ended up destroying the space shuttle program if you really get down to it. They did get back on it, after two and a half years. In the meantime the Defense Department was getting really nervous because they needed to get satellites up there. Before they finally cancelled the shuttle program a couple of years ago, like 90% of the flights were all military flights that were scheduled. I was in some of the meetings in Washington where the military is just — what are we going to do if we don't have a shuttle? Now you have private enterprise running flights and things like that.

§7.p2

There's one aside in that. I was testifying in an aircraft engine failure over here in the Cambridge courthouse, and they were asking me what I'd worked on and what I taught. I don't remember how it came up, but I basically said, well, things fail — for example, we have the space shuttle Challenger, and NASA knows there's going to be another failure sometime in the next 15 or 20 years and we just don't know when it's going to be. Guess what happened? The next morning the Columbia blew up. On Monday the other side settled the case, because the expert predicted it on the stand. I said, I — we don't know what it's going to be, but it just happened to be the next day. The insurance people were so nervous that some of the jurors were thinking I was a prophet who could never be wrong. So that was into that case. That's a true story. I was sort of surprised by all that.


§8. The structural welding code and the cost of standards [41:01]

§8.p1

I want to talk now about what the engineer's responsibility is. I didn't bring a copy of the structural welding code, but this is relevant to some other things we're going to go over, and design safety factors. The structural welding code — the steel welding code — is about an inch thick. It comes out about every three or four years. Costs about four hundred, five hundred dollars. There's a problem with codes and standards. Twenty years ago I could have bought that code for seventy-five dollars — basically not much more than the cost of printing.

§8.p2

I have a set of American Society for Testing and Materials x-ray radiographs. They actually x-rayed pieces of metal with different types of casting and welding defects. When I bought them in the mid 1990s, they were two or three hundred dollars apiece. Now they're two thousand dollars apiece — same stuff. There's a manual put together by the Defense Department and the Federal Aviation Administration that gives the properties of materials. It used to be called Mil Handbook 5 — a military handbook — but it was a joint thing between the FAA for commercial and the Defense Department for military aircraft. If anybody — Boeing, Sikorsky — was building an aircraft for the military, or building a commercial aircraft, they had to follow the properties in this manual. A lot of the stuff in the manual came from Boeing or Alcoa or US Steel, but it's just a collection.

§8.p3

They gave it a new name — it's called the MMPDS, Metallic Materials Properties Data Sheets. It's 11 volumes, about this thick. I paid 110 bucks for it. It was printed by the Government Printing Office. The Government Printing Office is not allowed to make a profit. At the time you could download a digital file for free, which I didn't do — I kick myself now. I was just involved in another matter involving aluminum and I was talking to some people, and I said, well, you can get this for free. At first I said it's now $700, because what the government has done is they've turned it over to a private company. I don't know how they're sharing the stuff and how they get around Congress's mandate that the government is not supposed to make money off these things.

§8.p4

It doesn't matter whether it's this particular document or other professional societies who write standards — they've all decided writing standards is a gold mine. The ASME boiler and pressure vessel code, the granddaddy code, is sixteen thousand dollars, and they come out with a new one every three years. If you're building pressure vessels, you've got to buy one, because you've got to know what the code says. It's written into law. You cannot build a pressure vessel and operate it anywhere in the United States without following the ASME boiler and pressure vessel code. Its status is written into law. They will not give you certification in the state you're in — each state regulates it — unless you can show that you met the ASME design requirements and fabrication requirements.

§8.p5

So there's lots of codes. The structural welding code's gotten up 400% in price. They used to redo these codes not that often. For a while they actually were trying to redo this one every year, and then people started screaming, this is ridiculous, the prices are getting out of hand. But in any case, this is written into code. This code is written into law in many cases — in Massachusetts, in New York City, most other states. This is the international standard for welding of steel.


§9. Engineer, contractor, owner: who is responsible? [45:11]

§9.p1

On page one, which is general requirements, 1.1 is the scope of the code and what it involves and the limitations on the code, and the definitions. First thing: the engineer shall be defined as a duly designated individual who acts for and on behalf of the owner on all matters within the scope of the code. It's not going alphabetically. The second one defined is the contractor. This is the person who's going to do the welding. The contractor shall be defined as any company, or the individual representing a company, responsible for the fabrication, erection, manufacturing, or welding in conformance with the provisions of this code. The owner shall be defined as the individual or company that exercises legal ownership of the product or structural assembly.

§9.p2

So the owner could be some financial group on Wall Street. The engineer is some guy who is going to have to put his stamp of approval on the design of this thing, and he's usually working for the owner. In fact, specifically he should always be working for the owner. In the back of the code, on page 415, it goes through a commentary — the C means commentary — for section 1, and the engineer definitions. "The code does not define the engineer in terms of education, professional registration, professional license, area of specialization, or other criterion. This code does not provide for a test of engineer's competence or ability. However, the assumption throughout this code as it relates to responsibilities and authorities assigned to the engineer is that the individual is competent and capable of executing these responsibilities."

§9.p3

Applicable building codes: you've got to build a building, you've got to use the code. The building contract will probably call out this code, may have requirements to be met by the engineer. The contractor inspector — most times the engineer works for the owner, and the owner is employing the contractor, but the engineer is basically supposed to be looking over the contractor's shoulder, doing the quality assurance function. For some applications of this code, one entity functions as both the engineer and contractor. In this code, this is referred to as the Original Equipment Manufacturer.

§9.p4

You know what a Butler building is? A prefabricated building. Butler in Kansas was the first big company to make prefabricated buildings. You send them: I want a two-story steel building that is 50 feet by 75 feet and I'm going to build it in this area of the country where the snow loads are such and such. Butler will design it for you, they will fabricate it, they'll ship it on trucks, and you can have it erected in a week. There are all kinds of Butler buildings. They're usually recognized as this cheap-looking metal building, and they have been designed down to just a very little bit of responsibility for the engineers.

§9.p5

This goes on for a couple of pages — engineer's responsibilities. Then you get to contractor's responsibilities: build it. All the rest of it is engineer's responsibility. So the engineer really has the bulk of the responsibility for the whole thing. We're going to talk about how the engineer's responsibility has changed over the last 20 years, and how we have come to new types of design criteria, now that we have the wonders of modern computers which sometimes can get you in trouble. So I will be in on Friday. Someone will be doing the next three days. I'll be in Canada.

Cases referenced

  • 1854 Broad Street cholera epidemic §4.p1

    Tufte's "successful" example of visual statistical thinking. John Snow plots cholera deaths on a London map, identifies the Broad Street pump as the cause, has the handle removed. Tom uses it to teach hypothesis-falsification, consideration of alternative explanations (people fleeing the neighborhood), and the role of outliers (the elderly woman who had pump water shipped to her).

  • Space Shuttle Challenger §5.p1

    Tufte's "unsuccessful" example. Solid rocket motor o-ring blow-by at low temperature. Tom teaches the engineering design of the SRM, the recoverable-vs-expendable architecture, and contrasts solid-fuel and liquid-fuel control. The launch politics (NASA pressure, $1.3B continuation contract, State of the Union, Christa McAuliffe) frame Boisjoly's failed warning. Feynman's o-ring-in-ice-water demonstration closes the case.

  • Bosco Lee Morton Thiokol O-ring engineering decision §6.p6

    The specific engineer (Tom calls him "Bosco Lee"; actual name Roger Boisjoly) who tried to stop the launch. The "take off your engineer hat and put on your manager's hat" line is the teaching beat. Tom returns in §7.p1 to note that Boisjoly resigned from Thiokol and went on the lecture circuit.

  • Space Shuttle Columbia §7.p2

    Tom's Cambridge courthouse anecdote. While testifying about an aircraft engine failure, he predicts NASA will lose another shuttle within 15–20 years; Columbia disintegrates the next morning. The opposing side settles the case. Tom uses it as a story about the strange theater of expert witness testimony.

  • Aircraft engine failure litigation (Cambridge courthouse) §7.p2

    The unnamed aircraft engine case Tom was testifying on when Columbia disintegrated. Used here as the setting for the Columbia coincidence, not for its engineering content.

  • National Ignition Facility (NIF) and the dual-use laser §5.p12

    Tom's neighbor's visit to Lawrence Livermore. Tom explains to her that NIF's stated mission (fusion energy) is cover for the real mission (simulating nuclear weapons effects after the test ban). Used as a parallel to the space-laser-on-the-shuttle case to teach how military programs are described to the public versus to Congress.

  • Space-based laser weapon and the cargo bay §5.p9

    Tom triangulates the classified laser-weapon design from three unclassified sources — Air Force 50 MW superconducting generator program, chemical laser efficiency at 10%, and Professor Bowen's one-meter potassium-chloride window forging at Lincoln Lab. Friend at Rockwell confirms the figure and the cargo-bay fit. Used to argue the shuttle's real purpose was military.

  • Lester Thurow lecture fees §2.p3

    $30,000-per-lecture in 1988 as benchmark for what top lecturers earn — and an argument that they don't use PowerPoint.

  • Edward Tufte seminar and book economics §1.p1

    The seminar ($500/day × 200 attendees = $100K/day), the four books on acid-free vellum, the handout pricing negotiation. Used as the source for the Napoleon's-march graph and the PowerPoint-Gettysburg satire.

  • Codes and standards as a "gold mine" §8.p1

  • Shuttle launch reliability §6.p7

  • Shuttle payload cost §5.p6

Layer 2 — cleanup edit
p1 00:00

In the back there is a handout which is from Edward Tufte. It's chapter 2 of one of his books. He has four books. Remember what he is a professor of at Yale, but he started getting interested in how to display information. And for example, this he considers his the best graph ever made, okay. I mean if you take his one-day seminar for, I think it's five hundred dollars, and they'll be very, oh yeah you have to understand it, but I won't spend a second with it. But he basically looks at lots of different ways to graph things, and sometimes you can graph things and actually have no value whatsoever to your graph or your table, and sometimes you can have all kinds of information and put it in one thing.

p2 01:00

Tufte has gotten into giving these seminars for industry. I took it over here at the Copley Plaza Hotel once. You pay five hundred dollars for one day, there's 200 people, so he's collecting a hundred thousand dollars for the day. Not a bad salary, although he does have to pay for the room, which probably cost him ten thousand. And when he's giving his seminar, he's made so much money off selling these books and giving these seminars, that he has, like he has Sir Francis Bacon's copy of Newton's Principia, okay. And you get to see the pages, because he will have underlings in white gloves carrying the book around showing you the page that you, he wants you to see. So he has a lot of these rare books and stuff because he likes to collect rare books. He has a lot of good ideas. He's also one of the most arrogant people I've ever met, but aside from that anyway.

p3 02:00

This is, you can't read the stuff up here, but it's in French anyway, unless you read French. But this is Napoleon's march from Paris to Moscow. Here's Moscow in French, and this should be Paris over here. And this is the width of this, is the size of his army going out. And then little pieces of the army went off to the side and got sick and went home or whatever. They finally get to Moscow, and here's the army coming back, and you can see the size of the army compared to what it was when it left, okay. So you can see why it's sort of a lot of information on this graph. And then there's the temperature plotted down here on the days that they were marching out. It was a cold winter and stuff. But for example, he considers that one of the most useful graphs, or most, I don't know, I can't remember, it's been fifteen years since I took his course.

p4 03:00

But he's written four books. He publishes the book himself. They're all on 100% vellum acid-free paper because he considers them to be a work of art that you will have forever, okay. But they're very inexpensive too. I think these books are like sixty bucks apiece, and we'll have the catalog with me right now. But however, this little handout cost $5. But it's on 100% rag paper, acid-free, so you will have it forever, okay. It will not deteriorate. I went up to him at the lunch break and I told him, I you know, professor Tufte, I'm a professor and I'd like to use this, but could I get a, if I buy in bulk like five hundred, could I get a discount? Well I guess you'll just Xerox it anyway, so I guess I'd better do it. So I got it for four bucks, okay. Well, it was still 1,600 or two thousand dollars or whatever that I paid for. And I've been using them for a few years for various classes.

p5 04:14

Anyway, so this is, we've been talking about decision making. Engineers have to make all kinds of decisions, trade-offs about different things. They also have to learn to communicate what they say, and so communication is supposed to be part of what I'm talking about in what we're doing here. So here it is: visual and statistical thinking, displays of evidence for making decisions, okay. And he has two examples in here. This is chapter two, one of the four books. One of them is, this is the first book is the visual display of numbers. The another one he calls the visual display of words. Another one of the visual display of verbs. No, not ones nouns, ones numbers, one that's nouns, ones verbs, and another, I don't remember what the fourth is. But he has a nice succinct way. He's a good teacher, okay.

p6 05:11

And here's one of my favorites, is the PowerPoint version, in auto content PowerPoint, of the Gettysburg dir— Gettysburg Address, okay. Have I shown you this before? Okay. This is the Gettysburg Address in PowerPoint, reduced to six PowerPoints. The organizational overview was, eighty-seven years ago versus now, okay. You graph that. Agenda met on the, back I guess I upload this up a little. The agenda met on a battlefield, it's a great battlefield, dedicate portion of field, it's fitting unfinished work, great. A review of key objectives and critical success factors: what makes nation unique, conceived in liberty, men are equal, shared vision, new birth of freedom, government of, or by the people. Not a, not on the agenda, his dedication consecrating, hallowing in a narrow sense to add a director, to add or detract, nor remember what we say. Summary: a new nation, civil war, dedicate field. He took a great piece of English literature and PowerPoint can reduce it to pure drivel, okay.

p7 06:29

I use this as, now that you're getting your PowerPoints ready for your presentations. PowerPoint is, it's a great crutch for people to present their outline, or to read from it. The best lecturers, the ones who make lots of money like Lester Thurow making thirty thousand dollars a lecture in 1988, or Tufte making a hundred thousand dollars for a day's worth teaching, they don't use PowerPoints. I mean I've talked to you about that before. You really want to have the audience focus on you and what you're saying, rather than up there looking and trying to figure out what the graph means. In fact I think even Tufte, who thinks that the graph of Napoleon's march to Moscow and back is one of the greatest graphs, he wouldn't expect it to be a useful thing to just put up. Because just like Catherine said, what's this, you know. I mean it's too complex. But once you actually get a little bit of an explanation, oh gee, you know, that's pretty neat, right. All of a sudden you have a visual image of how many people got lost and where, and okay. And then show you some of those lines where they were crossing a river that was frozen, there's something, and people fell in and froze and things like that anyway.

p8 07:54

But I want to take you through, and the scientific method. And we talked about the scientific method before. Do people remember the four key points of the scientific method? I wasn't that impressed with how many people rattled them off. I think you rattled them off, started to rattle them off. You first collect the data, okay. If you have a, the reference I gave you started with defining the problem, but I, so assuming you know what your problem is, you've got to collect data about it. You have to analyze it. Develop a hypothesis and test it. And only then can you draw conclusions.

p9 08:48

Now this has been around for thousands of years. I think Aristotle and others used it. But if you develop a hypothesis and test it and it doesn't work, you've got to go back and develop a new hypothesis, and you keep on doing this until you develop a hypothesis that cannot be refuted, okay. If you, you're trying to falsify your hypothesis. Your test should be designed to show that your hypothesis is false. If you do your testing and you cannot show it's false, then the conclusion is it must be true, okay. And there's a quote from somebody famous, I can't remember who, but when we tested something and the only thing that remains, no matter how improbable, must be the truth, okay, if we've eliminated all the other possibilities, okay. Maybe it's Sir Arthur Conan Doyle or someone, and that's good old Sherlock Holmes. Used deduction, right, to solve the problems.

p10 10:00

Now there's also induction. And the thing I gave you about three or four pages on the scientific method is the most complete description I've ever seen. It talks about confirmation bias, where people draw their conclusions in the beginning too early before they've laid out enough hypotheses to test, and they don't eliminate all the other possibilities. For example that I gave you, the thing out of Kahneman's book, where someone had looked at the data and they found kidney cancer was worst in the poor rural counties, right. And you can define that as the cause. It also turns out, if you look at the data another way, you find that kidney cancer was least in all those counties. So it was the worst and the least, depending on what county you're talking about. That was, as Kahneman said, that was the statistics of small numbers. If you're not averaging over a large enough cut list.

p11 10:56

There's expectation bias. You think that is the cause, and so you end up proving it's the cause. The confirmation bias, I often call it the reverse scientific method. You draw your conclusion, you throw out all data that doesn't fit, and you go backwards. And it's circular reasoning. You start with your conclusion and end up concluding the same thing. You're just going around in circles. You don't accept any data that doesn't fit, okay. So if you were going this way and then this way, you would not have that. If you avoid confirmation bias, you will test all the potential theories.

p12 11:40

So Tufte gives us two examples. One is John Snow, who in Lilith's, see what he says here, this is general for both of them, okay. What you know Snow, oh okay, well this John Snow was actually the anesthetist to Queen Victoria. Now why she needed her own anesthetist, have any time, do you ever give, been anesthetized? Anyway, so he was a medical doctor. Anyway, Tufte likes to point out that depending on how you display your data and analyze it, you may get better results than others. And he gives us a successful story and an unsuccessful story.

p13 12:26

And what happened was cholera broke out in the Broad Street area of central London in 1854. John Snow had investigated earlier epidemics, suspected the water from a community water pump was contaminated. Well, this is before we really had clean water. I mean the nineteenth century people were lucky to have clean water. People didn't understand bacteria in 1854. And he talks about what some of the theories were, that you contracted cholera through the air and things like that. But he had a hypothesis, okay. He had a good idea, and he was gonna use that hypothesis to test the data, and he was then going to evaluate things. And he goes through, and goes through the whole process of considering other things even after you think you've proven things.

p14 13:19

So down here at the bottom of this is a graph showing the death rate starting on August 31st. People started dying, peak the next day, and then it started to wane. Here's the cumulative death rate. 600 people died during the same epidemic of only a few days, about a week's time so far as that goes. And you've got some of these graphs if you picked up one of these things in the back. But what he did is he plotted on a map of London, and it turns out right here in the centre there's a little dot. See, it says pump, see pump, it says pump. It's not very clear cause it's small. This is the most deaths, okay. Each one of these represents a little line, and so the longer the bar, you can see the individual lines. This thing's not sharp enough to distinguish them. You can think of each one as a little coffin, okay. He has a picture later of little coffins.

p15 14:25

So he plotted, it turns out on September 7th he talked to the local town officials in that part of London, and he explained to them that he thought that it was the water that everybody was drinking from the community pump. And so they took the handle, the next day the town officials took the pump, the handle off, and the death rate decreased. Now you have to consider alternative explanations and contrary cases. For example, gee, it was already decreasing before he even did this, right? And so it goes through, and then explains, you have to say, well, why is that? Well maybe it's not the right hypothesis, if it already had started to decrease. But in fact what had happened, everyone was fleeing the neighborhood. After two days of, you know, a hundred and eighty people dying in one day, or whatever the number was, a lot of people decided to go visit their relatives in the country. Good idea, okay. And so therefore, you didn't have a constant population, and if you could have corrected for that population, you may have found that it was staying constant.

p16 15:37

And he gives a lot of anecdotal examples of, there was this one woman, elderly woman, and she loved the taste of the water from the Broad Street pump. And so once a week or something her nephew or son or something would go into London and she would have him bring back some water from the Broad Street pump. And he did, and she died the next day, even though she was a long way away from central London, okay. Yeah, or whatever, she was elderly. You know, if cholera may take two days, but if you're elderly may take only one day as you get definitely sick, okay. In any case, the stories in here, okay, I don't remember all the details, you can read it. I didn't, it's not something in my general knowledge. But he goes through a couple of pages here to talk about how they tracked down some of the outliers, okay. And they could all of a sudden, as you can explain the outliers, the ones that seem to disprove your hypothesis, and then you find that, oh, those have an explanation too. All you're doing is reinforcing your hypothesis, right.

p17 16:45

So far as that goes, so he goes on for several pages, we're not going to go through all that. He points out that you can plot it, plot the deaths on a daily basis, the same plot we showed you before, or you can plot it on a weekly basis. Sometimes when you aggregate things, in this case you still see the decrease, but sometimes when you start averaging things you can get really bad data. And actually he gives you another example from something else that shows, you've got a little peak here but if you average over a certain amount, actually averaging is something that electrical engineers, mechanical engineers do a lot in signal analysis. They may have a moving window, or they have a Kalman filter, they have different types of ways to filter the data to take the noise out. Sometimes you can take the real data out too if you filter out enough, you just get a straight line. And here's the little plot of Queen Victoria, John Snow, and a plot of coffins, okay. So that's, you can read the whole story yourself. It's actually good reading when you're doing nothing, okay.

p18 17:56

So the next one is a space shuttle Challenger disaster, which is, not one, 1986. And here's the space shuttle, and it has solid two solid rocket motors, it has one liquid oxygen liquid hydrogen tank, the center tank. That's expendable. The solid rocket mod— rocket motors are actually recoverable. They'd have a parachute, they'd eject them and fall into the ocean, and they'd recover it and reuse it. Whereas the main tank was not recoverable. It went further out into space and got burned up in the atmosphere. But the solid rocket motors were used, were reusable, designed to be reusable.

p19 18:46

And in order to be reusable, these things are so big you had to build them in modules, and you had to have o-ring seals, or, they had, you didn't have to have o-ring seals, you could have done something else I guess, but they had a double o-ring seal. And there's different pictures of this in the booklet. But they knew that, good old NASA has things down to fifty-eight point seven-eight seconds. Wow, good, okay, I'm glad they know it's that accurate to the millisecond. But you can see a little white spot at one of these o-ring seals as the solid fuel was coming out, okay. And the nice thing, well, if I know, the advantage of using a liquid fuel in the main tank and a solid in the, or the disadvantage of solid rocket motors? Yeah.

p20 19:48

Yep. Once you ignite it there's no way to stop it, okay. Can't put it out, because it actually is a solid fuel that contains its own oxidant and fuel mixed together. It doesn't need air, okay. Liquid oxygen, liquid hydrogen, you've got valves, and you, there, you've got a flowing liquid, you can shut the valve and all of a sudden you can throttle back your energy, okay. So the solid rocket motors, that's pretty blunt force rocketry. You light it and you hope for the best, okay. Whereas something that has, you know, that you can meter, or a liquid, has certain advantages. And obviously as you're getting up there and you're starting to maneuver and stuff, you'd like to change your power, and that's why they have a liquid-fueled rocket, okay.

p21 20:38

So they knew what had happened was blow-by the seals. And how many people know the story of how people analyzed it? You know the story of, well? [Student response] Yeah, I mean NASA basically shut down. They do that after every launch. Well, NASA basically shut down for about two and a half years. There were no shuttle flights for two and a half years. And what is the main purpose of the shuttle? You might know the main purpose of building the Space Shuttle? What do they tell the public, and what do they tell the hidden secret sessions of Congress? Right.

p22 21:27

It's a shuttle, just like the shuttle from here to New York, right. That's why they called it the shuttle. And at the time it cost ten thousand dollars a pound to put a payload, a pound of payload into orbit, okay. It still does, okay. The shuttle was advertised in the early 70s as it was going to drop the price by a factor of ten, and it would only cost a thousand dollars a pound, that's to get into orbit, okay. And people say, oh, we could colonize the moon. Yeah, twenty thousand dollars a pound, okay. That means, you know, it would pay to have someone go on a diet before they go on their flight, right.

p23 22:15

[Student] They had schedules. [Tom] And but as you said, the real reason, or the real justification, was military. They had to get payloads into orbit, they wanted to be able to get there and work on them, okay. Like the International Space Station. But they had military objects that were hush-hush, like space-based lasers. And it turns out, just by coincidence, Rockwell International had the contract for the space laser weapon. And there are various ways, you know, there's things that are classified. Unfortunately before I ever had a security clearance, I didn't have to worry about what I figured out, because I never learned it through classified means, right, because I didn't have a clearance.

p24 23:14

I had figured out through three different things that the power of a space-based laser weapon was 50 megawatts, okay. The Air Force was trying to get a, I'm sorry it was five megawatts, the Air Force was trying to get a 50 megawatt generator, superconducting generator, and I'd worked on superconductivity in the early 70s. And so kind of, the Air Force wanted a 50 megawatt generator, it's known that the lasers, the chemical lasers, were about 10 percent efficient. I can multiply 50 by 0.1 and come up with five megawatts as the laser power. Professor Bowen was working on one meter diameter laser windows for Lincoln Laboratory, forging these things out of potassium chloride. No one had ever tried to make very high purity, low imperfection, you know no dislocations or voids and stuff, things of potassium fluoride and potassium chloride and cesium chloride and stuff. No one had ever tried to do that before. But that's because the wavelength of the laser, this was transparent to that particular wavelength, the laser. You could look in the literature and you could find the damage threshold per square centimeter of laser power that you could go through it. You could multiply it by one meter, which is what he was trying to, no separate contract, you put the two together and you come up with five megawatts. And then there was something else, I don't remember what the other one was.

p25 24:39

So I asked a friend of mine who worked in the aerospace industry in California. I said, Charlie, do you know what the highest power, what the space-based laser's power would be? I figured it out and I told him my two methods. And he says, well, we have the main contract for that, so I really shouldn't tell you, but let me tell you that it's a pretty good estimate you've got. And what's more, it fits perfectly in the cargo bay of the shuttle, okay. This was like 1978 or something he was telling me this, okay.

p26 25:17

There are other things. It wasn't, I was just talking to, oh I was talking to a woman, lives around the corner, and she happened to, her son is working at Lawrence Livermore. He's working on supercomputers. And so they had a, I don't know something, but she went to visit Lawrence Livermore Lab. And Lawrence Livermore Lab, as you know, is one of our weapons labs. I visited there once, because I had a student doing a thesis, and he's still at Lawrence Livermore. And every building I went in had one of these little triangular temporary signs, and it would say "uncleared visitor today," which meant everybody had to keep their file cabinets locked while I was there, even though I had like three people escorting me everywhere I went. I was never any safer than, the only time I was less safe than that was when the KGB was following me in Moscow and Kiev, okay. And then I was safe as long as they didn't turn on me, right. But, you know, I had a police escort wherever I went, right. Well here at Lawrence Livermore, I had a security escort wherever I went.

p27 26:24

Anyway, so this woman around the street was, and she's a virtuoso violinist, okay. She's not technically inclined, other, her son is a computer scientist. She said, she started telling me about this big laser. I said, you mean NIF? Oh yeah, that's it. And they're gonna, they're doing fusion. And she actually, she's pretty bright, she asked the question, well how much power have you gotten out of this? And they tell her, this is nothing yet, okay, we haven't reached breakeven. Aren't you, spent all these billion dollars, have been working for fifty years? And then I explained to her, well, the real reason for the National Ignition Facility is to simulate nuclear weapons, because we can't do actual weapons tests anymore. So they built this five billion dollar laser, or set of lasers, so they can hit something small and they can implode it, get nuclear fusion, and test their weapons, the physics of the weapons. So a lot of things are sort of dual-use. They tell the public one thing, we're trying to make free energy by nuclear fusion, which they sort of are, but the main reason they're doing it is to simulate nuclear weapons effects.

p28 27:40

Anyway, one day before the flight the predicted temperature for the launch was 26 to 29 degrees Fahrenheit. The engineers who designed the rocket opposed the launching. [Student] Tell me more about that part of it. [Tom] They're in here, okay. Well, they sort of presented the wrong data. They pushed the data on blow-by, which is twice before they had some soot get past the o-rings, okay. Well it's 4,000 degrees on one side of the o-ring and it's, you know, whatever the ambient temperature is up there, minus 30 degrees in the atmosphere, on the other side of the joint. Here's a close-up of the joint. I mean, you've got the book so I'm not gonna blow all these up as nice as it could be. But basically here's the unstressed joint. When things get hot, there's gonna be, that's why they have to have two o-rings in here, and you can get blow-by one or two.

p29 29:19

And Tufte, who likes to be critical, said one of the problems with these presentations, and here are the handwritten presentations that Morton Thiokol, the designer of the solid rocket motor, who was a contractor to NASA. They didn't provide the names of the people who prepared the material, so no one was going to be responsible for this, okay. But it turns out the engineers at Morton Thiokol, which was out in Ogden, Utah somewhere, they had been successful in 24 flights. And the engineers went to the Thiokol managers and said, well we don't think you should fly tomorrow. And they'd already postponed it once. And do you know any of the politics about the flight? Well Thiokol was sort of in that position, that Thiokol had a 1.3 billion dollar continuation contract to continue for the next set of shuttles. And so the upper managers really wanted to please NASA.

p30 30:36

On the other part of the political spectrum, this was the one that Christa McAuliffe [Krista McAuliffe], the teacher from New Hampshire, and a few other people were on. And it just so happened Ronald Reagan was going to give his State of the Union that night, okay. And they were talking about potentially, once they're up there in orbit, actually he's talking to Christa McAuliffe during the State of the Union address. And, you know, everybody says, well we wouldn't, we didn't postpone it for that, we would have postponed it even if we would have missed that opportunity, yeah, in some people's minds. But in any case, it was still there.

p31 31:14

And what happened, there was this engineer named Boisjoly [Bosco Lee] who was one of the ones who helped prepare some of these slides and said, you can't do this, it's gonna, we have too high risk of failure. So the Thiokol engineers basically take that and they water it down. They call NASA in Huntsville or whatever and they say we don't recommend launching. And NASA called managers right back, or called back or faxed back, and say, this is days of faxes, and say, well you've never told us not to launch before. Well what does that have to do with any of this? Sort of a sunk cost, you know, problem. And there was all these politics going on. There's actually been whole books written about this, okay. And [Boisjoly] is, you know, they actually had a conference call scheduled, they were in the conference call, and it wasn't going very well for the decision not to launch. And NASA was trying to push Thiokol, and the Thiokol managers were trying to push their engineers into saying everything was okay. And so they finally decided to cut off the conference call and just have the Thiokol managers talk to the engineers. And one vice president of Thiokol told [Boisjoly], I think his name was Bob, hey, go take off your engineer hat and put on your manager's hat, okay. Well to me that's, you know, I told you engineering and management are basically the same. You can't take off one hat and put the other on. But they kind of pressured [Boisjoly]. Everybody's telling [Boisjoly], oh don't worry about it Bob, it's no problem, everything's fine.

p32 32:48

And they had, they plotted it wrong. They plotted blow-by and they didn't plot, they actually did have the data on there. This is one of the plots, and they have the temperature of the expected day. And they also had a, they had this is SRM 25, this is the 25th shuttle flight, okay. And they had had 24 successful shuttle launches, and NASA had done a multimillion-dollar study showing that the probability of having a failure on liftoff of a shuttle was one in 10,000, okay.

p33 33:30

Now it turns out in the investigation later, Richard Feynman, who was a Nobel laureate from Caltech, he'd started out at MIT, his career went downhill from there, he went to Princeton and then Caltech. But he graduated from MIT with a bachelor's degree in physics. But MIT, well, Physics Department will not admit their own students. They think there's plenty of other good physics departments, and that's true, okay, there are plenty of other good physics departments. So he went to Princeton, and he wrote books about the difference between Princeton and MIT. And then he went to Caltech, and he came up with quantum chromodynamics [quantum electrodynamics] and won the Nobel Prize. And he was sort of a spokesman for science. Anyway, at the hearing apparently, when NASA was saying, well this is a very rare event and we estimated the probability is 1 in 10,000 for a failure, and Feynman said, well I went back and looked at all the shots since World War 2, when we started rocketry, and we've had failures on 4 percent of our rocket flights. And this was the 25th shuttle flight, which is sort of coincidence, because anyone knows if it really was 4%, but we won't get into the statistics of that.

p34 34:48

But anyway, they had temperature of the o-ring, but they really were, this is history of temperatures, but the slide before was blow-by history, where you got a little soot past it. The o-ring history was a little different. They actually had o-rings, one of the o-rings basically burn up, and so you only have one o-ring left as the seal on some of these others. So two previous launches that had blow-by, and then they had some o-ring damage on SRM 15. SRM 15 was the next lowest temperature flight they had ever had: 53 degrees versus 29.

p35 35:33

And here on the next page Tufte gives two ways. You could do it in tabular form. If you know that temperature is your important variable and the coldest launch they'd ever had was 53 degrees, they had a damage index of 11 on the o-rings, they had only had two blow-by incidences, one at 53 or one at 75. But you can look here and you see there's sort of a cluster of o-ring damage here, and then there's a few that had some o-ring damage, they lost the rocket motor case in the ocean on that one. And you can plot it as a graph. And if you plot it as a graph, here's the index of damage, which is sort of going up like this. I mean, you know, you've got some data points here, but if I had to take all those, you kind of see this curve kind of going up like that, okay. And here's our temperature over here. If anyone had looked at this graph the day before the flight, they might have said, hmm, maybe we should look a little closer at o-ring damage. It becomes obvious in hindsight, which is part of what all this is. But what he's pointing out is you really do need to look at the data the right way, or you can miss the important parts of it.

p36 36:50

So here's Feynman, and he was testifying before Congress, and he went out and he asked for a cup of ice water. They thought it was cause he wanted to drink it, but in fact he had gone to the hardware store and he picked up a C-clamp, and he had a piece of rubber tubing, and he clamped it, put it in the ice water and showed how it was no longer, when he clamped it and put it in the ice water, got it cold, took it out, undid the clamp, you could see it didn't spring back to its original shape right away. There's no resiliency at lower temperatures for rubber. So that was how he demonstrated. Then Tufte talks about how even Feynman knew that was sort of Mickey Mouse experiment, but it still kind of showed the point.

p37 37:35

And here it talks about the difficulty for the rocket maker to deny the demands of its major client. NASA was really pushing for the go-ahead, okay. There was a possibility of televised conversation of Christa McAuliffe with President Reagan. And then Feynman said in the final report, "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." So, and that, some of the stuff that I have in the book by Kahneman, Thinking Fast and Slow, you know, you have your kind of intuitive thing. Basically Thiokol kind of intuitively noted we never tested these things at 29 degrees Fahrenheit. In fact, we thought we were going to be going in Florida, Cape Kennedy, Cape Canaveral, okay. In fact they usually were, but it was an unusually cold time at Cape Canaveral. I'm sure we probably lost a billion dollars worth of oranges that week, okay, but anyway.

p38 38:40

So there's a successful thing with John Snow, and unsuccessful one with the space shuttle. But the other thing about the space shuttle is, engineers, [Boisjoly] went on, well, he resigned from Morton Thiokol, and he went on the lecture circuit, and he used to say he wished he had stuck to his guns and told the managers that they were wrong, okay. But he took off his engineer's hat and put on the manager's hat, and it ended up sort of destroying the space shuttle program if you really get down to it. I mean they did get back on it, for, after two and a half years, the meantime the Defense Department is getting really nervous because they needed to get satellites up there. And I don't know if you know, before they finally cancelled the shuttle program a couple of years ago, like 90% of the flights were all military flights that were scheduled. Because the military, I was in some of the meetings in Washington where the military is just, what are we gonna do if we don't have a shuttle? Because, in a, well, now they come along and you have private enterprise running flights and things like that.

p39 39:52

I guess there's one aside in that. I was testifying in an aircraft engine failure over here in the Cambridge courthouse, and they were asking me what I'd worked on and what I taught. And I don't remember how it came up, but I basically said, well, things fail, and for example we have the space shuttle Challenger, and NASA knows there's gonna be another failure sometime in the next 15 or 20 years and we just don't know when it's gonna be. Guess what happened? The next morning the Columbia blew up. And on Monday the other side settled the case, cause the expert predicted it on the stand. I said, I, we don't know what it's gonna be, but it just happened was the next day. And the insurance people were so nervous that some of the jurors were thinking that I was a prophet who could never be wrong, okay. So that was into that case. That's a true story. I was sort of surprised by all that, but anyway. So as few other people.

p40 41:01

Anyway, I want to talk now about what the engineer's responsibility is. I didn't bring a copy of the structural welding code, but this is relevant to some other things we're gonna go over, and design safety factors and things like that. But the structural welding code is the steel welding code, which this came out of. It says structural welding code steel. It's about an inch thick. Comes out about every three or four years. Costs about four hundred, five hundred dollars. There's a problem with codes and standards. Back twenty years ago I could have bought that code for seventy-five dollars, basically not much more than the cost of printing.

p41 41:42

I have a set of American Society for Testing and Materials x-ray radiographs. So they actually x-rayed, that they took, of actually pieces of metal with different types of casting, welding defects. When I bought them in the mid 1990s, they were two or $300 apiece. Now they're two thousand dollars apiece, same stuff, okay. There's a manual put together by the Defense Department and the Federal Aviation Administration that gives the properties of materials. It used to be called Mil Standard, Mil Handbook 5, for a military handbook 5, but it was a joint thing between the Federal Aviation Administration for commercial and the Defense Department for military aircraft. And if anybody, Boeing, Sikorsky, anybody was building an aircraft for the military, or if they're building a commercial aircraft, they had to follow the properties in this manual. A lot of the stuff in the manual came from Boeing or Alcoa or US Steel or whatever, but it's just a collection like this.

p42 42:51

I bought it about six or seven, they came out, they gave it a new name, it's called the MMPDS or something, but Metallic Materials Properties Data Sheets or something like that. It's 11 volumes, it's about this thick. I paid 110 bucks for it. It was printed by the government printing office. Government printing office is not allowed to make a profit, okay. And at the time you could download, you get a digital file, which I didn't do, kick myself now, you'd get a digital file for free, okay. So I was just involved in another matter involved aluminum and I was talking to some people, and I said, well, you can get this for free. And at first I said, no, it's now $700, because what the government has done is they've turned it over to a private company. And I don't know how they're sharing the stuff and how they get around Congress's mandate that the government is not supposed to make money off these things.

p43 43:50

But doesn't matter whether it's this particular document or other professional societies who write standards, they've all decided writing standards is a gold mine. The ASME boiler and pressure vessel code, the granddaddy code, is sixteen thousand dollars, and they come out with a new one every three years, okay. If you're building pressure vessels, you've got to buy one, cause you've got to know what the code says. It's written into law. You cannot build a pressure vessel and operate it anywhere in the United States without following the ASME boiler and pressure vessel code. Its status is written into law. They will not give you certification in the state you're in, each state regulates it, unless you can show that you met the ASME design requirements and fabrication requirements.

p44 44:37

So there's lots of codes. The structural welding codes got up 400% in price. They used to redo these codes not that often. For a while they actually were trying to redo this one every year, and then people started screaming, this is ridiculous, okay, the prices are getting out of hand. But in any case, this is written into code. This code is written into law in many cases, in Massachusetts, in New York City, most other states. This is kind of the international standard for welding of steel.

p45 45:11

And on page one, which is general requirements, and one point one is the scope of the code and what it involves and the limitations on the code, the definitions. First thing is, the engineer shall be defined as a duly designated individual who acts for and in behalf of the owner on all matters within the scope of the code. Then those, it's not going alphabetically. The second one that it defines is the contractor. This is the person who's going to do the welding. Contractor shall be defined as any company, or that individual representing a company, responsible for the fabrication, erection, manufacturing, or welding in conformance with the provisions of this code. The owner shall be defined as the individual or company that exercises legal ownership of the product or structural assembly.

p46 46:05

So the owner could be some financial group on Wall Street. The engineer is some guy who is going to have to put his stamp of approval on the design of this thing, and he's usually working for the owner. In fact, specifically he should always be working for the owner. In the back of the code, it goes through on page 415, it goes through a commentary, the C means commentary, for section 1, and the engineer definitions. The code does not define the engineer in terms of education, professional registration, professional license, area of specialization, or other criterion. This code does not provide for a test of engineer's competence or ability. However, the assumption throughout this code as it relates to responsibilities and authorities assigned to the engineer is that the individual is competent and capable of executing these responsibilities, okay.

p47 47:04

Applicable building codes: you've got to build a building, you've got to use the code. The building contract will probably call out this code, may have requirements to be met by the engineer. These we were anyway, okay. The contractor inspector. There's, most times the engineer works for the owner, and the owner is employing the contractor, but the engineer is basically supposed to be looking over the contractor's shoulder and sort of doing the quality, the Quality Assurance function. For some applications of this code one entity functions as both the engineer and contractor. In this code this is referred to as the Original Equipment Manufacturer.

p48 47:49

So anyone, you know what a Butler building is? A prefabricated building. Butler in Kansas was the first big company to make prefabricated buildings. You send them, I want a two-story steel building that is 50 feet by 75 feet and I'm going to build it in this area of the country where the snow loads are such and such, blah. And Butler will design it for you, they will fabricate it, they'll ship it on trucks, and you can have it erected in a week, okay. There's all kinds of Butler buildings. They're usually recognized as this cheap-looking metal building, and they have been designed down to just a very little bit responsibility to the engineers, responsibilities.

p49 48:34

And this goes on for a couple of pages. I'm not gonna, actually, it keeps on going for all of this. It's engineer's responsibilities, okay. And then you get to contractor's responsibilities: build it, okay. But all the rest of it is engineer's responsibility. So the engineer really has the bulk of the responsibility for the whole thing. And we're going to talk about how the engineer's responsibility has changed over the last 20 years, and how we have come to new types of design criteria, now that we have the wonders of modern computers which sometimes can get you in trouble, okay. So I will be in on Friday. Someone will be doing the next three days. I'll be in Canada.