skip to main content
Home  /  Interviews  /  Thomas Heaton

Thomas Heaton

Thomas F. Heaton

Professor, Emeritus, Engineering Seismology;

Director (Ret.), Earthquake Engineering Research Laboratory

By David Zierler, Director of the Caltech Heritage Project

April 7, 20, 28, and May 10, 2022

DAVID ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Thursday, April 7, 2022. I'm delighted to be here with Professor Thomas Heaton. Tom, it is great to be with you. Thank you for joining me today.

THOMAS HEATON: Thank you for having me.

ZIERLER: To start, would you tell me your title and affiliations here at Caltech?

HEATON: I'm a Professor, Emeritus, of Engineering Seismology. And I have a joint position in geophysics and also in mechanical and civil engineering. And at the time of my retirement, I was the Director of the Earthquake Engineering Research Laboratory.

ZIERLER: When did you go emeritus?

HEATON: I went emeritus technically, I think, in fall of 2020.

ZIERLER: Right in the pandemic.

HEATON: Yes. Crazy timing, really.

ZIERLER: Crazy timing.

HEATON: I was kind of hoping to have some end-of-career things. We just had an event a month ago, it was the Heaton Symposium on the Future of Earthquake Engineering, and that was great fun. That had been put off various times, and fortunately, we moved to time it just when Caltech was reopening, so that was nice.

ZIERLER: Of course, going emeritus doesn't mean being fully retired. In what ways do you remain engaged in the field?

HEATON: I'm a co-PI on the Early Warning Project for the West Coast of the United States called Shake Alert. It's a collaborative with the US Geological Survey and a number of universities. For years, I was the PI, then when I went emeritus, I became a co-PI on that project. Probably will for about another year or so.

ZIERLER: For younger scholars in the field, as they look forward to the future of their career, what are some of the really exciting things happening that could occupy them for decades to come?

HEATON: That's a good question. There are just some really fundamental unanswered questions in the business. Earthquakes themselves, the physics of earthquakes, is still highly mysterious. Most people just get told the earth is locked up, increases strain over time until it gets to the breaking point, and then it has an earthquake, and everything goes back to where it was. That's kind of a very simple model of an earthquake. As it turns out, it's just so fantastically wrong. We always get this question, "Is it overdue? When's the next one?" We have no idea. For years, I gave a talk that I titled, "Why are Earthquakes so Gentle?" And that's, in my view, the real fundamental mystery of the earthquake problem. And people have tried to make earthquakes in the laboratory. They get some giant press, and they take rocks, and they confine them at the pressures that you would encounter 10 kilometers deep in the earth, and that's a confining pressure of about 3,000 atmospheres.

Basically, the pressure in the greatest depth of the Marianas Trench. It turns out if you take any kind of material, and you squeeze it together that hard, friction doesn't allow it to move. It takes enormous forces in the laboratory to make a piece of rock move at those pressures. And when it does move, it's a very violent thing in the laboratory. People take little, tiny, centimeter-sized samples, and they don't put their fingers next to the sample when it goes because it's very dangerous. And if you said that's what an earthquake looked like, and you scaled the whole thing up to the size of the earth, if it was like what we see in the laboratory, then the earthquakes would be just enormously violent. It wouldn't matter how you built your buildings, you'd be killed by the ground shaking, jamming your legs through your head.

Of course, earthquakes are powerful, but they're nothing like what we see in the laboratory. In a sense, this problem has been hanging around for 40 or 50 years now, and it's still not solved. At the end of the day, it has a lot to do with our understanding of the physics of the process, but it also has to do with how we design our buildings. Because people often say, "I'm designing for the worst shaking in 2,500 years," or some long number like that, as if we actually know how to make that calculation. The truth of the matter is, we're struggling with trying to understand the physics of the phenomenon in the first place. I think most laypeople and even most other scientists don't quite appreciate how unsolved a mystery earthquakes are. They're just a tremendous mystery.

ZIERLER: Based on what we know, do we at least have enough data so that there are candidate explanations for why earthquakes are so gentle?

HEATON: Well, of course, I have my own opinion. I think my opinions are probably close to the truth. But I'm not sure many of my colleagues are on board with me on this. In years to come, it'll be interesting to see how this finally gets solved. I'm sure we'll solve it eventually. But I think the solution is going to be quite interesting and rich in physics. It's a mechanical problem, and we're not talking quantum mechanics here, we're talking classical mechanics, in a sense. And there are two kinds of fields of classical mechanics that have come up. There's the standard linear classical mechanics that has been developed over the last three centuries, and that's a thoroughly solved problem. In classical mechanics, you can describe all the energies, all the motions going into the future, and it's very elegant, it takes a lot of training to learn how to do it, but it's a solved physics problem. But then, there's another part of classical mechanics, kind of non-classical mechanics, which has to do with when you turn up the amplitude of things.

In that case, most mechanical systems eventually transition into chaos, meaning that if you change the initial conditions in your mechanics problem by some tiny amount, then some years later, the system can act completely differently. It's sometimes called the butterfly effect. It came out of fluid dynamics, when people looked and said, "It's funny about our equations. If I just change things a little bit, my predictions of the weather into the future just completely change." And you see that all the time when people are trying to predict weather. They can predict a week out, but trying to predict a month out is kind of hopeless, and it's because it's a chaotic system. And I'm convinced the earthquake problem is a chaotic system as well. The problem is, numerically, those chaotic systems are extremely hard to solve. And if you really want to solve them, you've got to know all of the conditions in your problem.

You've got to know all the stresses and all the material properties, and then put them in some giant computer scheme. For us, in the earthquake business, all that physics is hidden by miles or kilometers of rock. We can't really see down deep in the earth. The only way we can tell what's going on is to listen to it with seismometers and look at how the surface is deforming over time. We've gotten pretty good at that, but to tell dynamically what's happening in these systems is really almost impossible. We have to be extremely clever to figure out from other approaches what's going on. And like I said at the beginning, one of the big questions has been, when we go to the laboratory at the tiny scale, we see very large stresses necessary to make a failure of rocks. When we infer what the failures are like for many other reasons, we think that things are failing at much smaller forces, much smaller stresses.

There was kind of a similar problem that showed up in materials science more than 100 years ago, and it had to do with an argument that went on between physicists and materials scientists and engineers. The physicists realized that to pluck an atom out of the lattice space in a crystal meant that you would have to move the atom by comparable to an atomic spacing to pull it out of its lattice point. They said the strain necessary to make the material deform was about one, the distance between the points. You had to move the atom about the same distance as the inner atom spacing. Then, they multiplied that strain times a material constant, a rigidity, and they ended up with enormous stresses of materials that would be necessary to make things move, sort of like 40 gigapascals, just stupid-big numbers. And the materials scientists said to the physicists, "What are you guys smoking? Why don't you measure this? You're off by a factor of at least a thousand. You're doing this the wrong way. There's something wrong."

They kind of fought with each other for a couple of decades, and eventually, they realized that the way the materials were deforming was, the atoms were jumping from one lattice site to another lattice site, and that would kind of leapfrog the process so the next atom would jump, and in materials, that's called dislocations. Dislocations would move through the crystal structure, and they would move at much less average force than the forces at the dislocation point. The strength of the material had an average number that was much smaller than the strength at the point of the actual failure. There was a size effect in this whole thing, but it's built into materials. Even today, materials scientists are trying to bridge the gap from the atomic scale up to the scale of things that we see, like cars or things like that. And it's an extremely challenging problem that really hasn't been solved yet. And I think it's basically the same problem we're dealing with in the earthquake problem. We're not talking about jumping atoms, we're talking about slip pulses, pulses of slip on the fault.

In earthquakes, the entire fault is not sliding at the same time. It's just kind of a solitary wave of slip that runs along the fault. At one point, those were referred to as Heaton pulses. Part of the reason I'm at Caltech is because I wrote a paper in 1990 that talked about slip pulses, evidence that earthquakes were comprised of slip pulses and what it meant for the physics of the earthquake process. And that paper was completely revolutionary and extremely controversial. In fact, I couldn't get it published in the standard journals. The reviewers just had too many reservations. And I withdrew it from the standard journals because they wanted me to change it, and I didn't want to change it. I sent it in to a more liberal journal, let's say. It was one where the chief editor was Don Anderson. He was the Seismo Lab director.

And I said to Don, "This is a really important paper, but these guys just don't want to listen to it." I showed him the reviews, and he said, "Yeah, you can publish it." It's quite a famous paper, and people still argue about it. But I think that's really the key to the problem. And since the time I wrote that paper, I spent quite a bit of my career trying to understand what the implications of that kind of failure are. And I'm convinced that it leads to chaos. And chaotic systems are just kind of a bear to deal with. The devil is in the details, and you don't know any of the details for these problems.

ZIERLER: To go back to this ideal idea that literally, we're just scratching the surface and can't go deep enough to see what's really going on, at least theoretically, if the challenges of drilling were not there, and we could go as deep as we needed to, do you think that would resolve all of these fundamental questions?

HEATON: Drilling would be good, but then you'd need it to have an earthquake. If you had the right drilling and the right earthquake, I think we could. I've been kind of also fascinated that this problem may be related to another problem in engineering, which is landslide problems. It turns out, most landslides, you go down and look at them, they're on fairly steep slopes. Looks like the coefficient of friction's maybe 0.46 on these steep slopes, something like that. The coefficient of friction's the normal force versus the lateral force. If you look at small landslides, and by small, I mean things that are only tens of meters or so, they're on a steep slope, then when the landslide happens, they end up in a pile at the bottom of the hill, and they basically don't go very far from the bottom of the hill, so they just head down the hill. Then, there are some other kinds of landslides, which don't happen very often, but much bigger landslides that go much deeper into the earth, maybe up to a kilometer.

When we see those really big landslides, they just slide forever. It looks like the landslide materials can end up kilometers away from the fronts of the mountains they came from. They look like they're frictionless when they're going. In fact, when Mount St. Helens, which was this big pile of rubble in a cone, failed for the eruption, the eruption was basically a giant landslide. We have photos of it, and time lapse photos show that the material looks like it was sitting on this cone, and suddenly, it was sliding downhill as if it were frictionless. The friction went from 0.5 to virtually 0 instantaneously all along this surface, which was the landslide surface, and the material ran all the way out for tens of kilometers away from the volcano. We know there are cases in the earth where deep in the earth, it suddenly transitions from very high friction to virtually frictionless. And people kind of argue about how that happens.

I like something called fluidization. When you have granular materials, if you get them sliding fast enough, they basically almost act as if they're a gas, but the gas molecules are actually pieces of rock. Of course, there has to be a lot of energy to keep that kind of system going. In materials physics, they talk about materials that are fluidized and materials that are jammed. I would say mostly in the earth, our materials are jammed. But when you get a pulse running through the material, you get a localized area of fluidized material, which has virtually no friction in it. That, to me, is an earthquake. It's this solitary wave going through the earth. Trying to predict how it's going to behave, you've got to understand how that friction's come about and the conditions that keep it there. You've got to understand the stress in the earth, and we don't understand the stress in the earth. In fact, in about 1995, one of the big projects in the United States was to drill a hole near the San Andreas Fault out near Cajon Pass to measure the stress on the San Andreas Fault.

They sent a lot of equipment down this 3.5-kilometer-deep well, not really deep, but deep enough to see interesting things. Rather embarrassingly, nobody ever really talks about it, but the answer was that when you looked at the orientation of the stress on this well, it indicated that the San Andreas is a left-lateral fault, which is clearly wrong. It's the wrong direction. It was a very embarrassing result, and people tried hard to explain it. I don't think they quite got it right. But the stress in the earth is probably extremely complicated. I think it's probably a fractal, and fractals are extremely complicated material states. They make the math a nightmare because the key element of a fractal is, every time you go to a finer scale, you turn up the magnification, you see more complexity. And if you're trying to use mathematics and differential equations to explain your system, if it's a fractal, life is really hard because you can't use calculus. If you remember taking calculus, it was, "What's the value as you take the limit as you go into an infinitesimal patch?"

In a true fractal, there is no limit. Every time you try to take a finer view, you get a different answer. It's not differentiable in any way. The standard equations don't work. In fact, there's a prize out there that's been out there for 30 years, where the fluid mechanicists realized that if they took the standard equations of fluid motion, Navier-Stokes equation, very simple equation, they had lots of solutions to this equation for smooth solutions. It's called laminar flow. Everything looks simple and easy. But if you turn up the velocity like in airplanes, and you make them go fast enough, everything eventually goes turbulent on you. And turbulent systems don't look anything like laminar-flow systems. It turns out, if you take that standard equation, Navier-Stokes equation, put it in the computer, and solve it numerically, you turn up the speeds in the system, eventually, it goes turbulent, just like in the real world, at least in the numerical solutions. But nobody's ever been able to do the mathematics to do that transition and derive those turbulent solutions analytically. There's a big prize out if you can do it, and nobody's ever done it.

To me, it just shows the richness of not-yet-understood physics that's out there in the chaotic world. We've gotten very good at solving the smooth world, the one where we can take all the derivatives, and those solutions are established science. Settled science, I'd call it. But chaotic systems, not at all, really. And they're all around us. Chaotic systems are in our economics, they turn out to be in epidemics, wars, things that really matter to human beings are all chaotic phenomena. Weather and climate are chaotic phenomena. Actually, this is an important issue. We keep saying, "The temperature's going to go up 0.5 degrees every decade," or, "The sea level's going to go up at this rate." And that kind of is describing a world where things are not truly chaotic. Truly chaotic things can just completely change state. Things can just get really haywire quickly in chaotic systems. And I think the climate people are not fully appreciating that yet.

ZIERLER: To go back to the paper and the trouble you had getting it published, what were some of the orthodoxies in the field that you were violating at the time?

HEATON: The prevailing theory at the time was something called characteristic earthquake, which was, suppose the last earthquake had five meters of slip, on average, on the fault. And plate tectonics in the area was working at a meter per century. Then, you'd say, "Well, the average time between earthquakes would be 500 years." And that's kind of elastic rebound theory. We're a little more sophisticated than just that, but not much. The idea that it was these pulses was really–people just hadn't thought of it that way. It turns out, for instance, that we have some average quantities that we know. The average slip on earthquakes compared with the length of ruptures turns out to be about a couple of parts in 105. If you took the average slip divided by the lengths of the total rupture, it'd be part in 105 kind of numbers. But in these pulses, the slip over the length of the pulse was more like a part in 103.

In the pulse idea, things were going on at spatial gradients that were two orders of magnitude faster than the average numbers in the earthquakes, and that really upset some people. They were used to think about all of these average numbers, and I said, "Actually, the average just comes from these other kinds of phenomena that are going on in the earthquake." I made the argument based largely on a bunch of observations that I had made with some of my colleagues. My PhD work back in the 70s was on trying to make computer simulations of the shaking in close to earthquakes. And I studied a number of fairly large earthquakes, San Fernando being probably the most important, and my colleagues and I made these models of how the slip was happening in the earthquakes. And I realized in the 80s that those models really required this idea of these pulses running through the material, and that that was completely inconsistent with the standard model of earthquakes at the time, which is sometimes called crack models. People in our business kind of borrowed the mathematics of people who studied the propagation of cracks in solids.

When I looked at the actual data that was coming in and the models that we needed to explain that data, the models were not crack-like, but they looked like these solitary-wave pulses. And solitary waves are a real nuisance to deal with. Everything's local in a solitary wave. All the physics happens locally and propagates through the material, whereas in the crack-like models, everything's kind of global, determined by the length of the rupture. All of the models before were, "I'm going to model a magnitude 7 earthquake, and the average length of a seven is 30 kilometers, so I'll make my model at M 7, assuming I've got a 30-kilometer rupture." In the pulse models, it says they all start as a point, then there's this pulse running, and sometimes it stops, and if it stops at 30 kilometers on average, it now belongs in the category of M7. But the 30 kilometers doesn't have anything to do with anything except that it just managed to stop at that point, and now it just gets thrown in the bin of M 7 earthquakes. The physics turns out to be very different.

ZIERLER: I'd like to ask some overall questions about your research throughout the years. First, why is engineering so important to you? What does that tell us about the things you're interested in?

HEATON: It's interesting, in the earthquake business, ultimately, we have to justify what we do to the rest of the world. When I was a graduate student, I guess the ultimate reason was that I had three children, which meant that money was a little dear at the time. When I was a graduate student, I did some consulting, and people don't pay you to just think and create crazy ideas, at least not when you're a graduate student. You get a graduate stipend, but nobody's going to pay you extra. But I actually did consulting for a couple of companies that were working on projects like liquified petroleum gas and nuclear power plants. And those were incredibly important problems for society. Many earth scientists say, "I'm working on these important earthquake problems because earthquakes are damaging to society."

Then, they go off and write some paper that's got nothing to do with society and really no connection back to how to make society safer. To be honest with you, I couldn't sell that. But just personally, it always kind of bugged me as being not really honest. If you are going to tell people that your work is going to make them more secure, probably it has to connect back through some other thing, like how things are built, early warning, how to protect yourself. Throughout my career, it's been especially satisfying to find interesting physics problems that also have implications for how we build our buildings and how we react to earthquakes because it just brings our work back to the people who pay for it.

ZIERLER: Over the years, how important has field work been for your research?

HEATON: That's a good question. I'm not sure whether I'm an observationalist or a theoretician. I'm somewhere in between. But we're trying to explain the natural world, and you have to look at the natural world to see what's going on. I like to hike, and anybody I hike with knows that I'm often out looking at the rocks, saying, "How did that happen? What was actually going on there?" It's really informed my brain a lot to actually go and see rocks, although no one would claim I'm expert at it. I don't actually write papers about it, but it informs my thinking. I actually like to look at buildings, too. What makes a good building, and what makes a bad building? How do you know the difference? And it takes quite a bit of experience and skill to learn that. I think having both of those skills is quite unusual, actually.

I know many engineers who know a lot of things about buildings but don't really pay much attention to the earth, and I know lots of earth scientists who know a lot about the earth but don't pay much attention to buildings. But that interface is incredibly important to all of us. Understanding which buildings will have the best overall performance for society is a really important problem, and it's particularly a pertinent question for Caltech because the Caltech campus is unlike any other set of buildings in the Western US that I'm aware of. Maybe there are some in Japan kind of like Caltech. But I don't think any other campus, at least, is designed like Caltech. Unfortunately, I think a lot of people on the Caltech campus don't quite understand that at the moment. And that's too bad because there's a view that if it's old, it must be bad, but that's just not true.

ZIERLER: Because social impact, the societal value of earthquake research, is so important to you, who have been some of the most important financial supporters of your research over the years, both in the public and private sectors?

HEATON: My PhD at Caltech ironically was supported by the US Air Force, which had a big grant for supporting the Seismological Laboratory, and they weren't especially interested in earthquake safety, they were interested in bombs. But after I graduated, I went to work for a year for Dames & Moore, which was a consulting company. I worked on practical problems. Then, I went to work for 16 years for the US Geological Survey, and that really was quite an important thing because the USGS said, "Our job is to provide scientific expertise to the rest of the country, to the government in particular, when they need some advice on some question with respect to safety and earthquakes." They go to the USGS and ask questions. "Our job is to do good science and give them well-thought-out answers that are not confused with lots of politics." The mantra when I was in the Survey was, "Go out and do good work. If it's good work, and you know it, don't worry about it. We've got your back. But if you're going to go out and get into all kinds of political things, we can't protect you." My background was, I tried to do science for the good of the country. And when I joined the Survey, we were told we were trying to predict earthquakes, and the more I worked on that problem, the more I concluded it was just a fool's errand. There's just no way we'd be able to do that. It led to the pulse paper, and the pulse paper basically says that earthquake prediction really is impossible.

ZIERLER: In all of your research, as you say, you're somewhere between theory and experiment, so on the theory side, what are the foundational theories that provide guidance for your research?

HEATON: There's standard continuum mechanics, which I learned as a graduate student at Caltech. Part of the reason I guess I'm in between engineering, and the science part was that the best people in continuum mechanics at Caltech were in the engineering department. It was called applied mechanics, and I discovered a guy, Jim Knowles. He was a professor of applied mechanics and just a truly brilliant man, outrageously good lecturer. I took every class he ever taught. I took six semesters of classes from Jim Knowles, and they were all just fantastic. I learned all my classical mechanics from Jim Knowles. Then, later, I learned other theory from my PhD advisor, Don Helmberger, who taught me a lot of theory. In fact, it's kind of a funny story, when I took Don Helmberger's class the first time, it was so foreign to me. I had no idea what problem we were even solving. I went through a whole semester of his class, and I don't think I really learned anything except I thought, "This is really weird stuff. I don't know what he's talking about."

One of my colleagues, Chuck Langston, and I we were looking for advisors, and he started working with Don Helmberger. He was a close colleague and having some success with his research. I said, "Wow, that's really interesting research." Chuck was able to explain it to me, and I understood what he was telling me. I said, "Maybe I should give this a try." I started working with Helmberger, and because of that, I decided to take his class a second time, hoping I would actually pick up what he was trying to tell us the first time. The second time through, at least I knew what problem we were solving, but it was still pretty opaque to me. Because I'd been through his class twice, and I was Helmberger's student, they made me the teaching assistant for his class, so I went through a third time. And by the end of the third time, I understood the problem, I knew what was going on. Half of my PhD thesis was to translate his class notes into something I could understand, and people used the appendices of my PhD thesis in other universities around the world to teach that theory because I was so determined to try and figure out what it was Helmberger was teaching us. That's where I picked up that theory.

Then, when I became a professor at Caltech, that was tough because I was in the engineering department, and there's a lot of teaching in the engineering department. The other professors came and said, "We want you to teach this engineering mechanics class," and I said, "I never took that class." They said, "Well, we teach that class here, you're a professor in engineering. Here, go for it." I was having to teach classes I had never taken, which turned out to be really tough, but it was incredibly transformative to me intellectually because strength of materials, a new way of looking at mechanics of materials as they fail, that was now on the books, things like plasticity and other things I hadn't seen before. It completely transformed how I did things. Furthermore, I started teaching something called dynamics, which is how things vibrate.

And the classical theory, this linear theory I mentioned, is solved, very elegant theory. As part of that, I started teaching nonlinear dynamics. Nonlinear dynamics had been around a long time, but if you turn up the amplitudes enough, it transitions into chaos. I had never really been that exposed to chaos until I started teaching it, and once I started teaching it, I realized, "Oh my God, that's the earthquake problem we're talking about here." To be honest, since that time, after having gone through teaching it, it's a hard subject, and I thought, "Wow, everybody in my business should learn this theory." But almost no one does. They're too busy, and they're already using the other theory, but it's the wrong theory.

ZIERLER: Given how deeply you've thought about chaos, is that to say that for earthquakes and earthquake prediction, the very notion of there being a cycle to earthquakes is wrong? That the idea that we're somehow, in California, waiting for the big one doesn't even make sense?

HEATON: I despise the term earthquake cycle. We use it, and it gives the view of the classical theory. We call it close to equilibrium. Equilibrium is a low energy state where there's not enough energy to make a big earthquake, then you slowly build up to get enough energy, you get there, and then it drops right back down. That's the earthquake cycle. In chaotic systems, they're often called in physics far-from-equilibrium systems, which means there's plenty of energy in the system for almost anything. When it's pulses, the pulses are just traveling through the material, and they travel until they stop. They run out of energy somehow, but they only run out of energy locally. It's not a global property. Predicting when they're going to stop is just virtually impossible with today's technology. The earthquake cycle just doesn't make any sense at all with pulses in chaotic systems.

ZIERLER: What have been some of the technological and instrumentation advances that have been relevant for your research?

HEATON: When I started in the business in '72, virtually all the records were recorded on pieces of paper or sometimes onto photographic film. It would have a light source going on photographic paper or a light source going onto a film. The thickness of the light compared with the dimension of the film or the piece of paper was maximum a part in a thousand. You could measure a number, 1, to the biggest number, 1,000, if you were very careful with a magnifying glass to make those measurements. And then, you'd have to measure it for all these different points in time, and it was extremely tedious to try to turn a paper record into something you could put into a computer. Along came digitizers, analog to digital converters. The earliest ones were eight bits, and eight bits is 2 to the 8, 64. It was no better. It was worse than the paper. But at least you could get a lot of points out of it. Furthermore, all of our things either went on paper, or almost all of our records that got sent anywhere were sent on telephone lines, and the range of amplitudes you could send on a telephone line was this part in 64, too.

You could only send numbers like 1 to 64. But it turns out from the earth, the numbers from the quietest sites, where you're looking at the smallest things, up to very strong shaking, are more like 40 million, so 1 to 40 million as opposed to 1 to 64. If you really want to go from things that are small all the way up to things that matter, things that are big, you need something that can record over eight orders of magnitude. And that has happened. It's amazing that it's happened. One of the things that happened first was, there was a neurotic Swiss mechanical engineer–all Swiss mechanical engineers are slightly neurotic–and he made a new kind of seismometer. The old seismometers were just a mass and a spring, and you'd measure the motion of the mass on the spring. If you get a motion more than a factor of 100,000 on that kind of system, eventually, the mass starts bumping into things, and then it's useless.

There's a limit of how big you can measure a motion with a spring and a mass, which we call dynamic range. This Swiss guy came up with a new system where, instead of measuring the motion of the mass, he measured the force necessary to keep the mass stationary. What it meant was, the mass didn't move, and all you were measuring was the electrical signal necessary to keep it from moving. We call that a force feedback system. It puts all the important parts of the seismometer into the electronics. It's no longer in the mechanical parts. And that allows the system to go to seven orders of magnitude by doing that. When he did it, nobody knew how to deal with it because there was no system that could record with kind of precision. Remember, paper could only get three orders of magnitude, and the first digitizers just couldn't handle it. But starting about 1990, there was a guy at Harvard, as his PhD work, constructed the first 24-bit digitizer, very high precision.

Hiroo Kanamori and I convinced Caltech to install one of those stations in Pasadena in 1990 or so. And that was a tremendous success because then we could measure with that station everything from the smallest to the biggest from very high frequency to low frequency, and everything was in a digital file. Huge numbers of pieces of paper that were very awkward to deal with were replaced by this computer system that magically covered everything. Hiroo Kanamori and I were both very excited about it. I was the scientist in charge of the Pasadena office of the USGS, and Hiroo Kanamori was the Seismo Lab director. We both realized that we needed to put out more of these instruments because it would really transform our field. Unfortunately, there was not much in the way of funding through the USGS, but Hiroo Kanamori was able to get the development department at Caltech to interest the Whittier Foundation in developing some of this, and they paid for 16 more of these stations to go out in Southern California. And that really changed everything.

Then, following the Northridge earthquake in 1994, there was money now available, and I was still with the USGS. The USGS and Caltech teamed up to put in something called the TriNet Project that was paid for by FEMA. Ultimately, we put out about 250 stations at that point. Later, when we got the Early Warning system funded in California, we've expanded the number of stations on the West Coast now to about 1,100 stations, and the business was just completely transformed by that. That was probably the most important observational change. Of course, there was another huge revolution that was going on in the geodesy world. I didn't affect that part of it. But geodesy was done when I started by surveyors. They had transits, they'd go to take a device on a tripod, and they'd measure angles to different peaks, then they'd measure distances with what's called a chain. They'd have guys walk from one place to another with a standard-length measuring line, and they measured up and down with something called a leveling line. You'd have a stadium rod, and the telescope would measure somewhere on that stadium rod.

We'd get up and down from that. It was an incredibly tedious project. Just to measure the elevation from one place to another could take years, literally. And then, along came all these satellite systems, ultimately global positioning satellite. Now, you can make those measurements in your car. It's just amazing. And that's completely transformed the way we measure changes in the motion of the earth. In the 1970s, people were using that ancient surveying technique and came up with this startling claim that the central part of the San Andreas Fault out near Palmdale had bulged up by about 25 centimeters over a period of ten years. They called it the Palmdale Bulge. I didn't call it that, but it was called that. It got a lot of press. Ultimately, it got some money for the program because people were afraid that we were about to have a big earthquake in Los Angeles. Later, when we got better surveying techniques, it was pretty clear all of that was just error in the surveying. But we don't talk about that.

ZIERLER: There's so much talk about the promise of AI in seismology. Where do you see that as a real possibility, and where's the hype?

HEATON: AI allows you to process a lot more data doing the kinds of things that you would normally do if you had the patience to do them. People don't have the patience to do these things. When I started, we used to have an army of housewives to read seismic data. Literally maybe 15. We'd call them timers, and they'd look at analog data and pick arrival times, and those arrival times would go into a catalog, and we got much more precision in how we located earthquakes and saw lots of structures. And now, we can do that same thing without an army of–they wouldn't be housewives today, I'm not sure who they'd be. But now, we can do it with machines. We can see a lot more precision. There are a lot of structures in there, but to be honest, even when we made the big change, having an army of timers, we saw structures, but we never really quite knew what they meant. Now, we're seeing more structures, and we still don't quite know what they mean.

We're seeing structures, and clearly the structures are kind of multi-scale. I'd say they're a fractal of some sort. And it was interesting in the early 90s when people started to see structures at different scales. They started to use the word fractals and said, "Oh, look, all these things are fractal." And many scientists were excited about that observation. Then, later, they stopped talking about it because they didn't know what to do with it. They didn't know what it meant, how to use it in some way. We're still at that place in AI. What we're really looking for is some brilliant intuition, probably from a human, that says, "This is how to interpret that kind of structure you're looking at." I hope there's somebody brilliant enough out there to do this that is thinking about this problem.

ZIERLER: With all of your affiliations and responsibilities around campus, throughout the years, has the Seismo Lab been your intellectual home, so to speak?

HEATON: It certainly was for the first half of my career. Then, when I joined Caltech and got this joint position, I had two offices, one at engineering, and one in geophysics. Earthquake Engineering at Caltech has been losing people for the last 30 years.

ZIERLER: What's to account for that?

HEATON: People think it's a solved problem. Caltech works on cutting-edge problems, and most people think earthquake engineering is solved. Unfortunately, t's just as unsolved as the earthquake problem in general.

ZIERLER: What's the source of the misapprehension? Why do people think it's solved when it's not?

HEATON: Part of it is because people who build cities would never tell you, "We hope this works." They say, "This building's safe." The people who build the buildings want to be able to tell the person who's paying for it, "You're buying a known quantity." For years, they said, "It meets the code. It's the best code." Then, people said, "Yeah, but what does that mean? I can think of ways to defeat it." But 20 years ago, the engineers changed tacks and said, "Let's do something they call performance-based engineering," which says, "Let's build to a standard where it won't collapse in a certain number of repeat-time years." Generally, that's 2,500 years. For most human beings that's a pretty good piece of time. The engineers are saying, "We know what 2,500 years ground motion looks like." That sounds like a solved problem to me.

If you tell somebody, "We know what 2,500 years looks like," it means, one, I'm not going to change my mind in 20 years. That'd be crazy to say, "We know 2,500 years, but I might change my mind." Furthermore, in all universities, but especially Caltech, we're competing with energy, with climate change, with epidemics. There's a lot of stuff going on in the world. And the earthquake engineers are out there saying, "We're building for 2,500 years." Well, what's left to do in their problem? Once you hear that–of course, it's total fiction, but there are people out there who are selling it. In fact, they sell it to Caltech even. Our current buildings, there are people telling people, "My design is designed for the 2,500-year earthquake." I think they're out of their minds.

ZIERLER: What does that tell us about the Seismo Lab that it was more central that your research in the first half of your career?

HEATON: When I became a professor, and I had to be in both, I had to teach engineering classes, which was a big challenge for me. I don't think any of my colleagues in the earth science world had to teach the engineering classes. And it took a lot of my attention to learn new theory. Furthermore, I ended up with quite a few graduate students who were in civil engineering. The graduate students were really good students. They were disciplined, smart, and I liked to work with them. It turned out, more of my students were in engineering than in seismology. I got really interested in the problem of what would happen to big buildings and large earthquakes, and that was in the engineering part, too. Frankly, the work with pulses, how they worked, that's a continuum mechanics problem.

And my engineering students were more skilled at those things. I just had more resources available to the problems I was interested in on the engineering side. But ironically, because I'm kind of an odd duck and willing to call a spade a spade, most of my engineering colleagues at other universities looked at my research and said, "What's Tom doing? I wish he'd get with the program. Seems like he's not really working with us." Because to be a success in the real engineering world, ultimately, you're not a success unless you build something. Engineers build things. Here I was, telling them, "Maybe you better think about what you're doing before you do this building." "We don't want to think about it, we want to build it."

ZIERLER: In all of your work studying vibrations, I wonder if you see it as a two-way street. In other words, how buildings vibrate tells us things about earthquakes and how earthquakes operate tells us about how buildings vibrate.

HEATON: Totally. And in the vibration part, there are two parts. One we call linear, where everything is controlled by the elastic stiffness of materials. It's a complicated, completely solved problem, like I said, and I learned a lot about seismology by looking at buildings. Because ultimately, seismology is about vibrations of the earth. We always do linear vibrations of the earth. When I taught how buildings vibrate, I discovered things we were doing in the earth science part that were just wrong. Of course, my earth science colleagues didn't want to hear that. I said, "You need to take an engineering class." There are a lot of things that happen in the earth science part, a lot of seismologists are looking for the strength variations of the interior of the earth by looking at the wave speed in different parts of the earth.

They say, "If it's a lower speed, it must be lower strength." And in the engineering world, you just would never do that because it just doesn't work. People can make the measurements in the engineering world, the yield strength, the speed the waves travel. I don't think there's any correlation at all. It's crazy to even think it would happen. I often gave my earth science colleagues a hard time and said, "Why do you think this works? There's no evidence for it." Actually, this is a little bit dangerous because I'm getting to be an older guy. Us older guys, it's easy to get to be a little bit grumpy. Who wants to be grumpy? But if you've been through 50 years of seeing people head down the wrong direction, it's pretty easy to get grumpy. But I don't want to be.

ZIERLER: I'm curious, your perspective on the idea that earthquakes are not cyclic, how does that influence your work on earthquake detection?

HEATON: We have an earthquake early warning problem. One part of the earthquake problem is to let you know that the waves are on their way and what it's going to be like when the waves get to you. That's a detection problem. How quickly can you tell what's going on with an earthquake and what's about to happen somewhere else? That's a really fascinating problem. If the earthquake's already over, then you've got to take the data you've got available at any given time and say what's going to happen somewhere else. If it's a really big earthquake, the earthquake may take a couple of minutes to play out, and you have to get the messages out even while the earthquake is still happening. And then, what do you tell people? Because you don't know, it's not over yet. You can only tell them, "Here's what it is up to this point." Of course, everything has to play out just within a couple of minutes, so there's not time to tell them very complicated things. And the question is, what kinds of information might be useful to people? Some of it has to be with how we control infrastructure like pipelines or how we would control trains and things like that. Other things would be telling humans how to get out of harm's way, or if you were in a high-rise building, telling you what you were about to experience, which would be pretty different from anything you'd see on the ground.

ZIERLER: To clarify, just because earthquakes behave chaotically doesn't mean it's impossible to develop networks and sensing systems that can detect them before they happen.

HEATON: Well, to detect things before they happen, we detect foreshocks. Maybe this could be a little confusing. When they happen, we can talk about the actual origin time at which the instability happens in the fault. That's what we call the origin time. Then, there's a time you feel it, which is sometime later, when the waves get to you from the origin, and it's not just at a point. The origin time is at a point, but the earthquake is actually spreading out on some fault. If we're going to tell you what's going to happen at your position later, we've got to know what's happening on that ruptured surface approximately and tell you, "The waves are on their way. Here's when they'll arrive, and here's how big they'll be when they get to you." That's our early warning problem. Then, you've got to make some decision what you're going to do about it. Are you going to try to run out of your building, get under furniture, shut down processes? When do you take different actions based on the information that's out there. And this is the perfect problem that's at the interface between engineering and science because you're trying to get this information from science and make a societal decision with it.

ZIERLER: Realistically, with the best possible scenario, what kind of detection system is feasible that would actually save lives?

HEATON: Well, we built a system that tells people that it's going to shake and to try to protect themselves. How many lives it will save, I can't predict for you.

ZIERLER: What's the lead time, best case scenario, with Shake Alert? How much time do people have?

HEATON: Best case would be a minute, a minute and a half.

ZIERLER: Then, of course, on the other end of that, there needs to be a reliable communication system that gets the word out.

HEATON: Yes. I think all of us are now thinking that's mostly our smartphones. Right now, there are systems out there that just tell you, "Earthquake coming. Drop, cover, and hold on." They don't tell you how strong it'll be or when it'll arrive. It just says, "You're going to feel shaking. Drop, cover, and hold on." Personally, I have always felt it's important to tell people, "It's going to shake in ten, nine, eight, seven…" And when you say it's going to shake strong in however long, I think it's also important to tell them, "Light shaking in ten, nine, eight–don't have a cow. Light shaking. You've been through this before. Don't get too excited."

ZIERLER: How much does the minute lead time say about the limitations in our detection technology, and how much does it say that maybe the earth itself doesn't know when it's going to shake?

HEATON: All earthquakes, as near as we can tell, start more or less similarly, whether it's a magnitude-1 or a magnitude-8. Within the first half a second, they all look the same, more or less. By the time you get up to about a magnitude-5, things start to change, as far as I'm concerned. A magnitude-5 is over in about a second. If you're 50 kilometers away, you're going to shake in 12 seconds. The earthquake was over long ago. It only took a second for the earthquake to happen. If the earthquake was a seven, it could take 15 seconds for the earthquake to play out. Right now, it takes us about five seconds to send out an alert. Usually, by the time we send out an alert, a five or less is already over. For earthquakes bigger than seven, probably not. The ones five to seven are in between.

ZIERLER: Some nomenclature questions. Crustal stress, what does that mean?

HEATON: Usually, we're talking about the shearing forces on faults. Most earthquakes on the West Coast are in the crust. The crust here is down to about 30 kilometers. And most earthquake shaking seems to be coming from 0 to 15 kilometers, maybe 20 kilometers. Most of the earthquake activity we talk about comes from the crust. There are cases of earthquakes that are beneath the crust in the mantle. But those had not been usually the really heavily destructive earthquakes. And this is another one of the mysteries in our business, where you can and can't have earthquakes. We have earthquakes sometimes as deep as 650 kilometers. Those are truly exotic, in a sense. They're obviously there, but there was a paper 100 years ago that proved that you couldn't have such earthquakes. Basically, the confining pressure is just so high that you could never get frictional slipping on a plane at that kind of depth. Well, Hiroo Kanamori, Don Anderson, and I wrote a paper that said if you get enough shear stress there, and it starts to deform, it'll melt the materials, and you'll get what's called a shear band in the material. I think that's probably what deep earthquakes are. But those are deep in the mantle. Of course, nobody can look at that stuff up close. And we've seen some big earthquakes that have been in the mantle that, if they happened in California, they'd probably be quite destructive. Historically, we haven't seen things deeper than 20 kilometers here in California. Is it possible we could ever see something deeper, down at 30, 40 kilometers? I don't know. I get to retire.

ZIERLER: Your work on base-isolated buildings, what is a base-isolated building?

HEATON: The notion is that if you have a building that is–if you put it floating on water, and the ground shook horizontally, the building wouldn't feel it. The building would just sit there, and the ground would move around it. You'd be in a boat. Of course, you can't build a building on a boat, so instead, people put layers of rubber biscuits, and they'd carry the vertical load of the building, but the biscuits themselves could shear, and they could shear up to about 25 centimeters, something like that. If you sheared them more than that, it would break the rubber. They designed the buildings so they could move relative to the base up to 25 centimeters, and they put the building in–the foundation would look like a swimming pool without any water in it so the building could move laterally inside this confined area. And if it moved more than that, the building would run into concrete walls, the sides of that swimming pool-like area. The first ones were here in about 1986.

There was something like the San Bernardino Juvenile Justice Center. I was on a field trip, and this was before I joined the Caltech faculty. I was with the USGS working on earthquakes. They took us out and showed us their pride and joy, this new base-isolated building. They took us down to the basement, and I looked at it, and I told the tour guide, "This can't work – there's no way. We're only two kilometers from the San Jacinto Fault. That fault can move five meters, and this thing runs into things in only 25 centimeters. You're just completely off the charts here. This just won't work." Of course, this was a tour. He had no idea who I was. And he said, "Who are you? Why do you say that?" I thought, "There's just some complete disconnect between what I know and what he knows."

And that sort of started me down the path. The base-isolation guys keep telling people, "If you isolate it, your problems are over." Everything works great unless you get enough displacement of the ground. If it's too big, then the building actually ends up impacting various kinds of quasi-rigid barriers, and the damage is worse than if you had just fixed it to the ground in the first place. For some kinds of applications, it might make sense. If you've got a historic building that was going to be damaged anyway, if you isolated it, it might make it through a number of earthquakes it'd otherwise be destroyed by. But I think it's kind of not a good idea to build a building that way from the beginning because there are much easier ways to make buildings earthquake resistant.

ZIERLER: Your work on strong ground-motion research. Is there such a thing as weak ground motion?

HEATON: Yeah, most seismologists work on motions that you can't feel. Like I said, the amplitude variation that seismologists deal with is a good ten orders of magnitude. I've especially been interested in the upper range of the shaking, the shaking you feel, or especially damaging ground motion. I'm really most interested in what's going to happen to one of our cities if we really get a big, direct hit. We've really only had one earthquake in the West Coast, and that was 1906. Nobody really knows how many fatalities, but San Francisco, the largest city in the Western US, was basically obliterated by the earthquake. No earthquake's come even close to doing that since then. And people sometimes think, "Well, it's so long ago, nothing like that could ever happen again." But we just haven't had an earthquake like that since. We've seen cases where big earthquakes have done some pretty nasty things to cities. The Japanese have had some really hard lessons there, although you can think of cases which would've been worse. Unless you get a real direct hit. For LA, our worst earthquakes have been San Fernando and Northridge, but they're not really under the city, and they were not really that big of earthquakes, 6.7s. If you bumped them up to magnitude-7 and put them under the city, I think they would've been completely different kinds of earthquakes that would've probably killed tens of thousands of people.

ZIERLER: Is it a useful distinction in your research in looking at oceans versus land in plate tectonics?

HEATON: In a sense, it is. The big sense is that people don't live on the ocean floor. Of course, that's where the tsunamis start. If you've got a subduction-like earthquake that's on the land, then you've just got more people exposed to that kind of shaking. There was an earthquake kind of like that, the 1999 Chi-Chi earthquake, a magnitude-7 earthquake in Taiwan. Very unusual subduction zone. It was basically all on land. They had a good seismic network, so we have very good records from it. Fortunately, the really strong shaking from that earthquake was in suburban areas with short, very stout houses. If you'd have put that earthquake under Taipei, it would've been a catastrophic event. Probably would've taken down Taipei 101.

ZIERLER: We talked about some of the orthodoxies earlier in your career. What are the big debates going on right now, and how have you contributed to them?

HEATON: I think earlier in my career, when people were trying to predict earthquakes, we had things like the Palmdale Bulge, we had some really fundamental and really strong debates about things. We used to have something they called the John Muir Geophysical Society, this ad hoc group, they'd get two different groups together, and they'd argue it out. It was like today's politics. We had different crowds that just took really strong positions. That seems to have disappeared from this business. Of course, the engineering business doesn't want to have debates because that's bad for business. They do their best to keep those debates down. And the earth scientists don't get involved in the building part because the earthquake engineers just say, "They don't know jack about buildings. Don't talk to them." The earth science part, there are still fundamental problems, but we've gotten old enough that we just don't fight with each other much anymore that I know of.

ZIERLER: Being at Caltech all these years, I'm curious if JPL has been an asset for your research at all.

HEATON: Not a tremendous amount. JPL's been focused on seismometers going to other planets. That's interesting. They were very fundamental in the geodesy business, but that's not actually where my research has been. I've been up to JPL a number of times, but not a lot. We put in 150 community seismic network stations on the campus, which is pretty interesting. But by and large, not a lot for me.

ZIERLER: And going the other way, have you served in an advisory capacity for us to try to understand tectonic plates on other planets, such as Mars, for example?

HEATON: I have not. Plate tectonics has not really been what I work on. I've mainly worked on the physics of the earthquake and what they mean for the built environment. To be honest, I'm often just very amazed by the kinds of mechanical models that earth scientists use for the earth. They make the earth entirely viscous. In the engineering world, people know what viscous is, but they hardly ever use it for anything. Plasticity is where everything is done in the engineering world. Plastic is how things bend and permanently deform. When I look at the earth, it's plastic. But earth scientists don't use that physics very much. Some do, but most don't. And when you make things plastic, it makes the mechanics much more difficult. Your research is much easier if you don't make it plastic.

ZIERLER: Heaven forbid, I hope it doesn't happen, but let's say that we do have another earthquake like 1906. First of all, on the built environment side, how much better prepared are we now than we were over a century ago?

HEATON: Most of the buildings in 1906 were short wooden buildings. And short wooden buildings have to carry the same load as any other kind of building. They carry all your books, or a meeting with 20 people in it. When people built wooden buildings, they had to be stiff enough that you could put those loads in them, and they wouldn't bend so much that it would crack all the walls every time you put in books or a bunch of people. And wooden houses or structures tend to be very robust because of that. They're light, and they're braced enough that they don't flex much when you put a load in them. If you get an earthquake, then they flex. You've got to patch all the walls. But you expected that in an earthquake. But because they don't crack when you put regular loads in them, they're very resilient in earthquakes. California's very lucky that most of our structures are low-rise, wooden structures that are very good in earthquakes, but they're really bad in fires.

Some people say, "1906 wasn't the earthquake, it was the fire." I don't know. But it was a disaster. We almost lost San Francisco again in the 1989 Loma Prieta Earthquake. They got some big fires going. Fortunately, there were no winds, and they were able to get it out. But a lot of people were worried we might've lost San Francisco again in 1989. But in the meantime, California's also built a number of tall frame buildings, which are buildings that have columns and beams. It used to be when people tried to build against earthquakes, they tried to make it as strong as possible. You'd look at it and say, "How much force would it take to actually shear this thing over?" They'd brace it, put in walls and things. They made things strong, but it turned out that when they did that, it made things very stiff. Because they put in lots of braces and walls. When you shake things in an earthquake, as you make it taller, if it's stiff, you end up with enormous forces at the base of your building, and you just can't make a tall building that's stiff and strong enough to go through earthquakes.

In the 50s and 60s, there was kind of a revolution where the engineers began to realize, "I can make it flexible instead. I'm going to give up on trying to make it so strong. I'll make it flexible so when the bottom moves, the top is kind of vibrating back and forth around it, but everything's not moving as a unit." When you make it flexible, it turns out all the shearing forces decrease in the structure. They really pushed towards flexibility starting in the 50s and 60s. And suddenly, when they did that, they discovered they could take height limitations out of the buildings, so they could make them kind of indefinitely tall by making them flexible, provided that the ground motion doesn't get too big. If you have a tall building, and I move the base enough, then the weight of the building is trying to rotate the base of the building. If it's vertical, it's all in the vertical. But if it gets out of vertical, then it puts a torque on the base of the building. And if it's flexible, there's a maximum angle that it can lean over and be stable to gravity.

And that maximum angle corresponds to displacement of the roof of most tall buildings. It'd be about two meters of the roof of the building relative to the base. In any kind of shaking, the motion of the roof is roughly twice the motion of the base in the earthquake, so any kind of ground motion that's more than about a meter, and tall buildings are in jeopardy of becoming unstable and collapsing. Most records people have seen that they're using to make these judgments are smaller than a meter. The ones that are bigger than a meter, they've removed from the dataset because they're oddballs. Of course, those are the most important records. For an earth scientist–of course they're meters.

That's how plate tectonics works. That's why there are mountains here and all these landforms came from big displacements of the ground, and they all happened in earthquakes. From an earth science perspective, if I said, "I'm telling you, the ground's not going to move more than a meter in a future earthquake," an earth scientist would say, "That's crazy." That's not the way it's talked about. Right now, in the earthquake engineering business, the process is largely secret. Everything's done in proprietary reports. Earth scientists don't know what the engineers are doing. We're not allowed to know what the earthquake engineers are doing. I've been privy to that because I spent half my career trying to learn it.

ZIERLER: Is part of the challenge that the things that the things that you're interested in really need to be understood in a multidisciplinary environment, and so much of the academic side is siloed?

HEATON: I would say that's true. Maybe I should send you the PowerPoint for my talk I gave at my retirement symposium, which is basically about these issues. I can send you that PowerPoint, and maybe you can use some of that.

ZIERLER: Have you worked specifically to try to break down some of those silos, to get different specialists talking with each other?

HEATON: Well, I have, in a sense. But to be honest, I'm really interested in the technical parts of all these things. I'm not very interested in management things. And to actually run symposiums and things like that, I kind of detest that. It just seems hard. It's interesting, when Caltech hired me, they said, "We want you to be a bridge between engineering and earth science." I said, "OK," because I didn't know. But when I went into engineering, I discovered there are no bridges that can be built that are that long. It's just too big a gulf. I said, "Maybe I need to change my focus. Maybe I'm a boat that goes between these two." Now that I'm retired, I've decided I'm probably an island.

ZIERLER: Last question for today. Because all of the things that are important to you require effective government coordination, what have you learned about interface at the local, state, and federal level in order to get these things through?

HEATON: I have learned there are people in the political world who are honest actors for public good, and then there are other people who just don't want to be bothered. The former group is the one that makes everything happen, and they're very rare. But there are people out there that really care about the public good, and they're looking for answers from the technical community. Some of those people actually are solid thinkers and ask good questions. But lots of people are just going through the motions. Earthquakes are just easy to handle when you don't have them. Lots of people, it's out of sight, out of mind. But some people are thinking, "Wow, there's something out there." Eric Garcetti is one of those people who's paid attention to this problem, and he's communicated well through Lucy Jones, who's sort of in this interface stuff, too.

It'll be a change to lose Garcetti because we've had a bunch of mayors who haven't paid much attention. Frankly, we have to be careful what we tell them because if we get the attention of a politician who really makes a difference, if we don't give them good advice, they're not going to listen to us ever again. Like I said at the beginning, we need to be careful to give them advice about our fields. I've got to tell you, I'm a little upset with Caltech lately as an institution. Somehow, we can't figure out whether we're about science or social justice. And people don't care what we think about social justice. That's not what we're here for. If we start to get all over the world in all kinds of controversies about social justice, we're not going to have the trust of people for our science. We need it, and they need us to do it. I'm not sure what the heck we're doing.

ZIERLER: Stay in our lane is what you're saying.

HEATON: Yeah, we need to. I understand why people feel strongly about these things, but I think they're losing sight. That's why Millikan is out of favor. He was a great scientist, but he didn't stay in his lane. We should've learned a lesson from him.

ZIERLER: That's interesting. On that note, I'm very excited, we'll pick up next time and go all the way back to the beginning, learn about your family and your childhood. We'll go from there.

[End of Recording]

ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Wednesday, April 20, 2022. I'm delighted to be back with Professor Thomas Heaton. Tom, great to be with you again. Thanks for joining me.

HEATON: Thanks. Good morning.

ZIERLER: In our first conversation, we did an excellent tour of your overall science and the engineering, policy, the administration. We touched on it all. Today, I'd like to go all the way back to the beginning and learn about your family first. Let's start with your parents. Where were they from?

HEATON: My parents are from Missourah.

ZIERLER: That means they're really from Missourah. [Laugh]

HEATON: They're from St. Joseph, Missouri. My dad was an only child in a very small family. His family had set up the First Bank of St. Joseph in Missouri. And his grandfather was also in the banking business. Unfortunately, he died at an early age. I never got to meet him. My dad joined the Navy at the end of World War II. He never saw service, he was an aeronautical engineer at the beginning, and then he worked for, I think, McDonnell Aircraft in St. Louis for a while and decided that was a dead-end job. He went into school on the GI bill and got his mathematics degree at the University of Kansas. I was born while he was at Kansas. Then, he moved to Bloomington, Indiana to get a PhD in mathematics. After he got his PhD, he joined the faculty at Rutgers University in New Jersey. I basically spent most of my youth in New Jersey from age 5 to age 17. Once I left New Jersey, I never really went back.

ZIERLER: What campus was your father on at Rutgers?

HEATON: He was on the New Brunswick campus, which later became the Piscataway campus, which is where our house was. We lived in one of the first developments in Piscataway, which was kind of a rural agricultural area, and our house was a bunch of developmental houses, acre-lots just went along a street that cut into a wooded area, so I spent a lot of my early life out in the woods, climbing trees.

ZIERLER: Did your father give you a sense of what it was like to be a professor? Did he involve you in his career at all?

HEATON: Yes, he did. I remember when I was a teenager, I would be listening to him. He'd be upstairs in his bedroom, grading final exams. I'd hear out of the upper bedroom, him mumbling, "Idiot. Stupid. Nitwit." [Laugh] That's my first remembrance of my father with mathematics. We talked about science, and I was science oriented when I was young. And when I went to high school, I did really well in the sciences. Reasonably well in mathematics, but not tremendously good. It turns out that both my father and I are almost certainly dyslexic. In the time I was a teenager, and when he was a professional, the word didn't really exist, maybe some psychologists knew about it, but virtually no one ever knew about it. It was just obvious that he was an extremely absent-minded person. He lost his keys, lost his papers or whatever. Because he just didn't keep track of things. He was almost as bad a speller as I was. As I child, I think it was pretty obvious to people who were around me that I was bright, but when it came to getting good grades in elementary school, I wasn't that bright because I didn't remember anything. A lot of elementary-school learning is about memory work, and I was bad at that. I never really understood it at the time. I just knew it was hard.

It was hard for me to learn to read, and spelling was impossible. But somehow, I made it through. I went to Rutgers Preparatory School, which was a private school that was quasi-connected to Rutgers. There were some good academic children there. I don't know whether I stood out, but I was not a great student when I was there. It was only when I got to physics in college that I suddenly started to do well. I discovered, when I got to college, that studying for exams didn't help me at all. More than anything, it tended to confuse me because I tried to remember things, and my memory was so poor that trying to jam things into my memory was useless. It's interesting that later in life, I learned how to use the dyslexia to my advantage. I didn't really intend to, but because I knew I had a poor memory, I got to choose which things to remember. It's funny, you go through life, and you get this enormous amount of information, and it's all up there somewhere. Can you get to the stuff that you want to up there? For me, it was easier because I'd learned a lot of tricks for how to remember things I really wanted to remember. If I saw something and at the time, I thought it was important, I would use those tricks to say, "Put this in your brain because you want to remember this, it's important." I got to choose the things I remembered, which is kind of an advantage if you're a scientist because you don't want all that irrelevant stuff.

ZIERLER: Growing up, even if you were not academically inclined, were you interested in science, even as a boy?

HEATON: I was. I was always interested in science. But as a boy, I was kind of confused, in a sense. I wanted to be popular. My parents thought I was bright, and they somehow engineered it so I skipped kindergarten. I was the youngest child in my class going through high school, which meant that I was the smallest person in the class, not especially mature, I don't think, but I was small. And people thought, "He's smart, but why isn't he doing better at tests?" Of course, as all young men go, I wanted to catch the eye of the young ladies. But being the smallest was not to my advantage. [Laugh] It was a little frustrating, I've got to admit. But it was a great surprise that later in life, at least one female was attracted.

ZIERLER: You only need one. [Laugh] When it was time to start thinking about college in high school, in terms of your grades, how far away you wanted to travel, what was in range for you to apply to?

HEATON: It's interesting, at that time, in high school, they always had some sort of career counselor to try to steer you to where your career would be. And career counselors in high school typically would want to steer you to where they had already been. They'd steer you away from things they hadn't seen. And my counselors were from small private arts colleges like Gettysburg College. I didn't want to go to Gettysburg. We used to vacation in the summer in Maine, which is a beautiful place. I applied to Bates College in Maine, and I got in there. We visited, and it looked nice, but it turns out you can't really tell anything from a college visit except what it looks like. Once I got there, it was a tremendous shock what it was like to be with a bunch of other young, immature freshmen on a college campus. It was a zoo. [Laugh] Literally, they hazed all the freshmen, all the upperclassmen spent half their life in a drunken stupor.

It just really was unappealing to me. I was not a party person. I spent two years in Maine, and of course, the main thing about Maine is, the summer's beautiful, but the winter is forever. And it's a really tough, cold winter in Maine, and I had never spent a winter in the middle of Maine. It once got down to 40 degrees below 0. That's seriously cold. About two years in, I knew I was interested in science, but the Bates Science program was kind of mediocre. It was very strong in the liberal arts, but that wasn't really my interest. I decided that it was a good idea to make a switch in the middle. That's when I went to Indiana University. And that turned out to have been a really good move for me.

ZIERLER: Of all places, why Indiana?

HEATON: Well, my dad got his PhD at Indiana, and he had nice things to say about it. I actually had some very early childhood memories from when I was in Indiana. They have a very good physics department, and they have a good music department. I've always been very musically inclined as well as science. I chose it, and it turned out to have been just perfect because it was such a big school that you could get as good an education as you could stand at a place like that because there were honors level courses in everything. And they had a good physics department. And the physics instructors were good, better than I could get in Maine. I started out as a chemistry major because I was good in chemistry in high school. I did well up until I hit organic chemistry. If you know anything about organic chemistry, it's a lot of memory work. That was just torture for me. I said, "I can't stay in chemistry. This is horrible." That's when I switched to physics, and I did well. It just was like rolling off a log. It was really natural. I was basically a straight-A student by my senior year. And they didn't care whether I could spell or not. It was irrelevant.

ZIERLER: In making the switch, or even before college, was the draft something you needed to contend with?

HEATON: Yes, of course. I had a college deferment. This was the late 60s, going into the early 70s. My college deferment was going to be up in 1972, when I graduated. And we were still in the War. In those days, we had a lottery system, where they would, depending on your year of birth, they'd pick a number out of a big drum. They put 365 days in that drum, and all of us would watch the lottery draw to figure out what our number was going to be. And the low numbers got drafted. I was drawn as number 14, which meant that my post-graduate career was determined to be going to Vietnam. By that time, I had pretty much decided that was not a moral war and was something I would like to avoid. In fact, I was thinking whether or not I should have to relocate to Canada or something to avoid going to Vietnam because that was just a bad idea. But as it turned out, I kind of lucked out in the same way that Bill Clinton did.

We didn't know each other, of course, but in 1971, there was a time when the United States had several hundred thousand troops in Vietnam, and it was costing a fortune to support that war. In '71, the secretary of defense made a speech where he said it looked like the US would be decreasing their investment in the Vietnam War in '71. A number of us recognized that probably meant that they would not be drafting people the first quarter of '72. And the rules were that you only had to be eligible for one quarter of a year. If I dropped my deferment, so I was eligible the first quarter of '72, either they'd draft me right away, or else if they didn't draft me, I would have escaped because I would've given my eligibility. I dropped my deferment December 31 of '71. They didn't start drafting again until the summer of '72, so I'd done my exposure. I escaped the draft. At the same time, I met Norma Hart, who's my current wife. My life just completely changed in that winter of '72. I found the love of my life, I escaped Vietnam, and I got accepted for graduate work at Caltech.

And it's interesting, I was a physics major, and I had done some work as an assistant to a graduate student in physics at Indiana, and at that time, physics was all about nuclear physics. After Sputnik, the United States put enormous resources into nuclear physics because we wanted to make sure we could keep up with the Russians. Turned out by '72, we had so many nuclear weapons that the US just didn't need any more at all. The funding of nuclear physics research had really fallen off. I got into a bunch of physics schools for graduate work, but three of them sent me letters and said, "We're accepting you, but please understand that none of our graduates can get any jobs. There doesn't seem to be any future in this field." This is from the schools. They said, "Maybe you should consider some other field like geophysics or biophysics." I thought, "Well, I'll look." I liked geology, so I applied to geophysics at Caltech. I got in, and I came out to visit Pasadena in March. Pasadena's beautiful in March, it's spring. And the campus was just fantastic. I was seduced.

ZIERLER: As an undergraduate, what exposure did you have to geophysics and seismology?

HEATON: A little bit. I was a physics major, but I always liked geology, so I had taken some geology classes at Bates and also at Indiana University. I worked in a geology lab for a while, and then when I got to Indiana, I was in the geophysics classes at Indiana as well, and I did well in those. Some of my recommendations came from some of my professors in Indiana Geophysics, who remained reasonably good colleagues for years after I left. I had a geophysics connection at Indiana, but geophysics at Indiana meant oil exploration or mineral exploration. There are two parts to geophysics. One is finding resources in the earth, and the other is earthquakes. I knew nothing about the earthquakes. When I got into Caltech, it was just geophysics. I didn't really understand what it was until I got there. Then, I discovered it was earthquakes. It turned out in 1972, the earthquake field was just wide open.

It was sort of the beginning of being able to really do science in geophysics. It was a tiny field, and people didn't even know yet what the problems were to be explored. I got in at a time when there were lots of important problems to be solved, which, in a sense, was ideal because I just love big problems. It was fortuitous. Life is like that. I always told my students that, "Life is just a series of doorways, and you get to decide, 'Should I walk through this doorway or keep walking and do something else?'" And education is the key that allows you to open those doorways, education and experience. And if you're lucky, you go through the doorways that really allow you to find your destiny.

ZIERLER: Were there any professors at Indiana who steered you toward Caltech?

HEATON: No, they didn't know anything about Caltech. It's interesting, Caltech was such a tiny school, it was virtually unknown to the rest of the United States at that time. Only later after I got here did I realize what a really special place it was. But I didn't really appreciate that until I got here.

ZIERLER: Just to clarify, it was specifically the Seismo Lab that attracted you to Caltech?

HEATON: That's where my background was, geophysics. If I'd have gotten into physics at Caltech, which I highly doubt because physics at Caltech would've turned out to have been really high-level at that time, I'm not sure I would've survived, to be honest. When I got to Caltech, the geophysics department, in order to have broader education of the students, required us to take a class in the physics program, mathematics of physics, Physics 129, which was kind of the weed-them-out course for the physics grad students to test their mettle. To be honest, it was a horribly tough class full of a bunch of seemingly unrelated mathematical problems with very little insight into how you would actually solve them. It was really a difficult class. I didn't learn much in that class except that I was glad I wasn't in the physics program.

ZIERLER: When you got to campus, was the Seismo Lab still in the mansion off-campus?

HEATON: Oh, yeah, it was really fantastic. The mansion off campus, some of our classes were there, but most classes were on campus. My office was in the mansion, and it was just a fantastic place. It had its own private tennis court, and at lunchtime, we'd go down and play tennis on this court. Of course, we'd need to shower afterwards, and we'd walk up through the director's office, and the shower was one of these walk-in showers the size of a small bedroom. It was a very informal place. Don Anderson was the director of the Seismo Lab, and he was in his early 30s. He was a very informal kind of guy. Very creative scientist. But he didn't like formality. And typically, in the morning, we would talk about various science issues in the Seismo Lab coffee break, which typically ran to be about an hour long. The students and the professors would go down and talk about what their latest research was about. A lot of us, that's where we learned a lot of the most important parts of our research, in this coffee hour. It was held in the basement of the mansion. It was actually in the room with the boiler in it. There was an ancient, threadbare couch that was in there that people would sit at, and they'd sit on various concrete piers down there. Incredibly informal place. Interestingly enough, that couch got moved out of the old Seismo Lab, and it got recovered, and it's still in the South Mudd building. Amazing.

ZIERLER: Besides the inconvenience of having to go back and forth from the Lab to campus, what was good about being sequestered away in the mansion? What was useful in terms of your development and education?

HEATON: The Seismo Lab was a professional place. It provided information to Southern California about earthquakes. We'd just had the San Fernando earthquake, and there was a lot of interest in earthquakes in Southern California. And traditionally, the Seismo Lab had been the source of information about earthquakes since 1930-ish. After the 1906 earthquake, there was the seismographic station at Berkeley was established, and they hired scientists and set up seismometers to study earthquakes in Northern California. Then, starting in about 1916, there was a paper by Harry Wood, who was a geologist working out of Hawaii, who wrote a paper and said, "There are earthquakes in Southern California as well."

And he proposed, in this paper, setting up a similar organization to what was set up in Berkeley, and he proposed a network of some 16 seismometer stations across Southern California. He continued to work on that idea until it was funded by the Carnegie Institute, the same ones who did Mount Wilson Observatory. The original Seismo Lab was an endeavor of the Carnegie Institute, set up in the San Rafael Hills because they wanted to have a place to run seismometers where it was quiet and the seismometers were on rock, and they could even put them into little tunnels. The San Rafael Hills were the best choice at that time. Campus was all soil, not a good place for the seismometers. Wood was basically the first de facto director of the Seismo Lab, although he was never actually called that, to my recollection. In about 1932 or so was when Caltech was really expanding under not-to-be-mentioned Robert Millikan, who decided that Caltech having good geology, and especially an earthquake program, would be important. He basically managed an unfriendly takeover of the Seismo Lab. The Seismo Lab was brought under the wing of the emergent Caltech, and the Carnegie Institute kind of dropped out of the picture.

Then, they brought in Beno Gutenberg, a well-known seismologist from Europe. He was from a wealthy family in Europe that had a factory that produced soap. He was well-set-up in Europe, but somehow, they convinced him to make this change and come to this strange place, Pasadena, out in the middle of nowhere, and develop a geophysics program through the Seismological Laboratory. At the same time, Charles Richter had just graduated in physics from Caltech, and they brought Richter on as kind of a scientific assistant. But Richter was very interested in earthquakes. He really centered his career on it. Gutenberg was a much broader thinker. He thought about earthquakes, the structure of the earth, plate tectonics. He was a very broad, creative thinker. He's kind of the real hero at the Seismo Lab. The other guy they brought in was Hugo Benioff, who was basically a mechanical engineer, and he designed many of the instruments that were used to record the motions of the earth. The three of them were very different people. I don't think they were actually ever even friends with each other. I have many stories from some of the original people at the Lab of lots of intrigue and wild stories. I can't really repeat them.

ZIERLER: How did you feel connected to the founding of the Seismo Lab? Who was around that made that generational connection for you?

HEATON: Because we were the center of information for Southern California, whenever there was an earthquake, the press would call in and say, "People felt this. What was it? What does it mean?" The Seismo Lab was not running 24 hours a day, and there was no formal press office, so it was whoever picked up the phone got to be the person who would provide the information to the LA Times, TV, or whoever. For many years, if Charlie Richter was anywhere close, it was him. If there was an earthquake, he would go into his office, take the phone, and put it in his lap. And he'd respond to any request, and all the public things went to Richter, so he got a lot of notoriety because of that. But by the time I got to Caltech, Richter had retired and was not in good health. There were a variety of people who would answer information, and some of them were staff at Caltech, but often, it was the graduate students at Caltech. Especially since we worked late at night, there were always graduate students in the Seismo Lab.

Because of that, we were trained how to interpret the incoming seismograms, how to locate the earthquakes and assign a magnitude, and we were also expected to represent Caltech to the rest of the world usually on the telephone or occasionally on television. But because of that, we were part of a group of people who were kind of apprentice Charlie Richters to provide the information. That's completely changed now because the USGS has really taken over that role, and they provide a lot of information. These days, we wouldn't let a graduate student anywhere near the press. But when I was young, we were expected to talk to the press, and we were expected to talk with intelligence to the press because we were representing Caltech. That really was important to me, that I was learning to be an apprentice providing a service to everybody.

ZIERLER: As you said earlier, the field was wide open when you got to the Seismo Lab. What were the big debates at the time? What were people talking about, what were the different schools of thought?

HEATON: Probably the biggest problem at the time was that in the early 70s, especially at Lamont-Doherty at Columbia, they were working on a couple revolutionary problems. One, the plate tectonics revolution was important, and that was just getting going. People were realizing, "Oh, yeah, the plates are moving around," but the science behind how plates were moving around, what the plates were, and exactly how they were moving around hadn't been worked out at all.

ZIERLER: Had you appreciated that the Seismo Lab was the central contributor to the plate tectonics discussion at that point?

HEATON: In the 50s, Beno Gutenberg wrote some papers where it was pretty clear that he understood there was plate tectonics, and he talked about the elastoplastic Earth. This is prior to the plate tectonic revolution. But by the time the plate tectonic revolution came around at Columbia, at Woods Hole, Beno Gutenberg had died, so he never got to see that part of it. And it was ongoing when I got to Caltech. But it wasn't a big part of Caltech, the plate tectonic revolution, although some of the students at Caltech wrote their PhD theses on describing how the plates were moving. A very important paper was written by Tom Jordan and Bernard Minster about doing a mathematical inversion for all plate tectonic data to describe how the earth was deforming. In that sense, Caltech was there, but we weren't really the leaders in the field.

At the same time, there was another thing that was ongoing that came out of MIT and Columbia suggesting that there were measurable precursors prior to important earthquakes. And that started the search for earthquake prediction. In the 1970s, when I was a graduate student, many researchers were looking for precursors to earthquakes, including data from prior to the San Fernando earthquake. And that was a really hot problem, not only amongst scientists, but also with the press as well. The problem got to be so important that it led to the formation of a program called the National Earthquake Hazard Program, which provided a huge stimulus of money into this problem of earthquake prediction. As it turns out, most of the science that was done in there was very marginal. If there's anything we learned from that program, it's that we can't predict earthquakes. A lot of my career is about why we can't predict earthquakes, what about them makes them unpredictable.

ZIERLER: But to clarify, at the time, there was much more optimism that the science would get to a place where we could predict earthquakes.

HEATON: That's exactly right. When I graduated and got my PhD, my first position with the USGS was in the earthquake prediction program, and eventually, I became the USGS coordinator for earthquake prediction in Southern California. But I became very skeptical of the science that was being done about that. But when I first joined Caltech, there was a lot of optimism. I worked on several problems, but one was that there were a number of cracks seen in the area around Pasadena, specifically along the Raymond Fault. It runs mostly east-west from Arcadia out near the LA Arboretum and extends west to the Huntington Art Gallery. That hill is the Raymond Fault. It also runs all the way to the power plant in Pasadena. And there were a number of fractures of roads that were seen along that fault, and there was a question of whether this could somehow be a precursor to an earthquake in Pasadena.

In my first year, I was assigned putting instruments out along this fault and monitoring changes along the fault. And it never really came to anything, to be honest, but I learned a lot about instruments. Then, the next problem I was interested in was tidal triggering of earthquakes. Are there times when the pull of the sun and the moon accelerate failure process. Can you say you're more likely to get an earthquake when they're pulling in the proper direction? That was a pretty serious effort. I made a computer program that would calculate the stress on various faults, and then I could test, when earthquakes happened, whether the tidal stresses had been in the direction of the failure of the plane. And that was fairly new in the days that I did that. And I looked at a fairly large collection of earthquakes, and I found that the situation looked kind of confused. But if I separated the earthquakes into different bins, like whether they were shallow or not, or whether they were strike slip or thrust earthquakes, it made a difference. And I found that of the earthquakes I looked at, the thrust earthquakes that were shallow seemed to respond to tidal forces in the way that you'd expect, so I wrote a paper and said, "There seems to be an effect on this class of earthquakes, and it looks like it's statistically significant."

I did some statistics on it, but I was a very young man then and still pretty naive. After I wrote that paper, I continued to collect data, and about eight years later, I took the new data that I had and repeated the experiment I had done in the first paper, and I discovered that I couldn't repeat my results. I had to write another paper that said, "The previous paper of Heaton is not repeatable. It's incorrect. Disregard it." It was pretty painful, but it taught me a simple but important lesson. People have known this forever. There are lies, there are damned lies, and there's statistics. It's so easy to fool yourself as a scientist. If you have real, honest-to-God science, it's repeatable, and it's repeatable in a blind way. That's why in the medical world, people require there to be double-blind tests, so you can't know whether you've gotten the medicine or not to know about the result.

You can't change the criteria for how you make a hypothesis after you've seen the data. Seeing the data allows you to have insight, but if you want to do statistical significance, you better do it with data you've never seen. Otherwise, you'll be biased. And this is a huge problem in the earth sciences because there's only one earth. There's only one dataset. Redoing experiments is almost impossible in the earthquake business because the earthquakes make themselves, and you've got to wait a long time for new data to come in. Earth scientists routinely calculate statistical significance of data that they've already seen. Typically, they're just fooling themselves. They make a good story, but they don't make good science.

ZIERLER: What were the most important instruments at the Seismo Lab when you were a grad student?

HEATON: Of course, the seismometers were the most important things. We had a wide range of seismometers that were designed especially by Hugo Benioff, and then we also had a set of very important seismometers that were from the original vision of Harry Wood in the 20s called Wood-Anderson seismometers, and those are the seismometers that were used for the basis of Richter's magnitude scales. John Anderson was an astronomer in the 20s, and he worked for Carnegie Laboratory on Santa Barbara Street here in Pasadena. He worked on Mount Wilson. As an astronomer, he worked on optical systems. The original Wood-Anderson seismometer was a very simple, elegant device. It was a wire with a mirror on it, and it was slightly off-center so when the ground would move, the mirror would twist on the wire, and the wire would be the restoring force. You'd shine a light into the mirror, and as it deflected, it would send the light off onto a photographic paper and turn it into a seismogram. It turned out to be extremely simple to understand but very effective.

That system gave us stability of operation over about 70 years. I could look at a record from 70 years earlier and know what the motion of the ground was that produced that record. It turns out that's not so easy. You've got lots of people out there designing seismometers. How do you calibrate what it means to have a record on a piece of paper in terms of turning it back into the motion of the ground that made that record. That's a little bit of an art form, and some of us learned that. The real genius at that these days is Hiroo Kanamori. He's a little bit older than I am. None of us are getting younger. And there are only a few of us around still who know how those old records work and how to interpret them. I always used to teach my class that records are important, that's where you need to go to find the information, but you've got to know how to interpret it. I have a section in my class notes about how to do this. But after the mid-90s, everything changed to becoming computerized and electronic. It's definitely becoming a lost art.

ZIERLER: On that point, what did computers look like, and what were they used for when you were a grad student?

HEATON: When I first came to Caltech, my computer, I've still got it in my drawer out here, was a K&E Slide Rule. That was the easiest way to do things. Then, Caltech had a big Marchant Calculator. It was a special kind of adding machine, totally mechanical, but it could also do division and square roots. When I first got to Caltech, there was one analog computer that had a bunch of tubes in it. No transistors, all tubes. And that computer spoke its own machine language, and it was used exclusively for locating earthquakes from–you'd input the various arrival times at different stations, and there was only one person at the Seismo Lab who knew how to run that Bendix analog computer to create our catalog of earthquakes. That was at the Seismo Lab. Then, the year I was there, they first came out with the Hewlett-Packard HP-35.

It was a handheld calculator. In those days, they were $1,000, which today would be about $12,000 for a handheld calculator that could do sines, cosines, things like that. That calculator was so valuable, you had to sign out for it, and people would line up to use it for their research. On campus, there was, of course, a computer center. At that time, it was an IBM 360 computer that had 640 kilobytes of total memory, and in those days, that was just what was on the computer. There was no extended memory or anything like that. You had to do all your problems in 640 kilobytes. We'd all go in to do our calculations on campus, you'd submit a deck of punch cards to be run, and you'd get a printout of your calculation and your cards back, and if there was a mistake, you'd have to redo it. It was a very time-consuming process, but that's what it took to do calculations in those days.

ZIERLER: Did you do much field work as a grad student? Or was all the data coming to you?

HEATON: The work I did on the Raymond Fault was field work, and we had field trips where basically the entire division would go places. We went to Imperial Valley. I actually assisted some of the technicians at Caltech in putting in instruments after things like earthquake swarms or aftershock sequences. There was one technician in particular, Ralph Gilman, who had been there at Caltech since 1935. He kind of took me under his wing and taught me about seismic instruments. He's the source of all my inside information about the intrigue at the Seismo Lab, but unfortunately, I really can't divulge that stuff. There are some wild stories that Ralph told me. Ralph was a wonderful man with a very dry sense of humor and good understanding of what he was working on. I really learned a lot from him. In some ways, I learned as much from him as I did from the professors.

ZIERLER: What was the process of determining who your thesis advisor would be?

HEATON: When I got to the Seismo Lab, we had to, in a year's time, in the fall of our second year, go through an examination process. It's basically the qualifier exams. They were called propositions at the time. We had to present four ideas, and have done some research on those ideas, and talk them through well enough to present them to a committee of the faculty for a typically three-hour oral exam. This was an extremely stressful period. A young person going before a bunch of senior professors with four original ideas you're trying to claim you know something about. That's a lot. These days, I think they only require two. To do that, you have to go talk to some professors about these ideas ahead of time to make sure they weren't just completely nuts. Up until that time, we'd only been assigned to professors to help them with things they wanted done. But of these four ideas, eventually, one should turn into your PhD. I did four of these ideas, and it turned out none of those ideas had anything to do with my PhD.

I was a little anomalous in that case, but I had taken a class my first year, advanced seismology, from Don Helmberger, who was my eventual advisor, and it was a really obscure class to me. Don had a knack for making things really hard to understand. And by the time I got through that class, I just didn't understand what he was doing. I thought, "Well, I'm not going to do that." But one of my colleagues, Chuck Langston, had not only taken his class but was doing some research with him. Chuck was a really close friend, and we talked about what he was working on. I realized, "He's doing some really good stuff there. Looks really promising and has a good future." After I saw what Chuck was doing, I proposed to Don Helmberger that I start to work with him on a separate but related problem, the theory behind how the ground moves close to earthquakes. And Don said okay, he was a very accommodating man.

Because I did that, I decided to take his class a second time to sit in, to make sure I finally started to understand what he was talking about. After the second time through, I had a reasonable idea of what he was doing. Then, because I'd been in the class twice, and I was then his student, they asked me to be his teaching assistant. I went through it a third time, and by the end of the third time, I had a good understanding of what was going on. As part of my thesis, I took the class notes, and I rewrote them in a way that would make them more understandable to other people. People used my class notes in other universities to teach that theory for a number of years. The theory's hardly ever used anymore because it's been superseded by numerical methods.

ZIERLER: For your research, where were you more focused on the theory, and where was it about the experimentation and observation?

HEATON: In my business, I was interested in how the ground moves in damaging earthquakes. If you are in the business of putting out instruments and waiting for damaging earthquakes, you'll get old before you get the data. Earthquakes come when they want to. I basically used existing data, then I used theory to explain why those existing data were the way they were. At the time I did that, most of the people using the data about shaking in close to earthquakes were in the engineering department, and in those days, they looked at those records, and they just looked like a bunch of squiggles. They didn't know why they looked the way they did, they just characterized, "Here's the amplitude, here's the duration, here's the spectral content." But they didn't try to understand why they looked the way they did, which I was interested in.

When I started working on this, I was using that data, and I came to some places where I said, "This data doesn't seem to make any sense with my emerging understanding of how this works, but it would make sense if they mislabeled the data, mistook up for down or one component for another component." I wrote to the engineers, the people who had the data, and I said, "Are you sure this is the right orientation and stuff?" And they checked, and they were so shocked that I could tell that there was an error in the data because to them, it had always just looked random. And here's some young person who writes in and says, "I think you made an error." "How did he know that?" Of course, what I was studying was, "Here's the earthquake. What should the ground-shaking look like because of that earthquake?" That was sort of the focus of my PhD work, what makes the earthquake ground motion look the way it does.

ZIERLER: How did your thesis research, as we were discussing previously, contribute to some of the broader debates going on in the field at that time?

HEATON: There's a whole school of thought in seismology where, when people look at seismic data, they don't actually look at the seismograms, the squiggles, themselves. Instead, they take the seismogram, and they take a Fourier transform of it, which means they decompose it into sines and cosines of different frequencies, and then they look at the average spectral shape of the earthquake. And that was a very big field in the 70s, looking at earthquake spectra. By the way, that was pioneered by Jim Brune who was a professor in the Seismo Lab and left about 1970. But I wasn't looking so much at the spectra, I was trying to understand individual squiggles in the records, and I was under the tutelage of Don Helmberger, who was interested in all those squiggles. There was kind of two different sides of seismology. One was this spectral approach, and the other we called the time-domain approach. And those two groups often came to different conclusions about how earthquakes worked. There was a lot of friction about that between the groups over the years, and there still is, to be honest. But I like to think the guys who looked at the squiggles mostly won this war, but these wars kind of go forever.

ZIERLER: What was Helmberger like as a mentor? How closely did you work with him?

HEATON: I would see him roughly a couple times a week, and I would go into his office. I really liked the man, very patient and intelligent man. He was a very young guy, too. When I got to Caltech, he was probably 32, but he looked like he was 20. He looked like a grad student. He just loved games, a good tennis player, good athlete. We used to play football with him, tennis. But he was a very bright guy, too. Of course, I remember I couldn't understand him for a couple years. Finally, I did. Don had a lot of students because his students basically recruited all the other students over the years and said, "You ought to take a look at what Helmberger's doing. It's got a good future to it." That's the way a lot of the students went. And then, they worked with Don, and once they got to know him, they really liked working with him. But he had this very odd habit I never really truly understood.

We'd be talking in his office, having some discussion, and then it's like, boom, locked up. Just everything stopped. Don would be looking out the window for three minutes. Then, suddenly, it would start up exactly where it stopped. It was like you were in a time warp. And every student was completely unnerved by this experience. Different students had different interpretations of what was going on in his brain while this sudden hiatus occurred. I discovered nothing was going on because nothing ever seemed to have changed from before to the after part of these lapses. It was just a really strange thing that I got used to. Some people never got used to it. But we're all kind of strange in our own ways.

ZIERLER: In our previous discussion, we engaged in a very good overview of some of the orthodoxies that have changed over the years. With that in mind, what aspects of your thesis research have aged well, and what have not?

HEATON: Actually, I have gone back to the thesis at various times to see whether what I did was good or not. I've got to say, the thesis didn't make mistakes. I'm very proud of that thesis. But of course, it's become obsolete, in a sense. The part I said about the theory where I explained Helmberger's techniques, nobody uses that technique anymore because at the time we were doing it, it allowed us to solve numerical problems on the computers of the day, which were very limited computers. We have much bigger computers today to solve the same problems with today's numerical techniques. But with the computers of that time, to solve those problems, you needed to be very clever about what parts of the problem you relied on the computer for and what parts of the problem you relied on traditional mathematics. What I wrote in there was good, but nobody uses it anymore because it's just too much work to do it that way.

ZIERLER: It's not that it's outmoded, it's just labor intensive.

HEATON: What I did was labor intensive. Now, computers are so fast that you don't need to do that anymore. Just let the computer loose.

ZIERLER: Is there anything lost in outsourcing this to the computer?

HEATON: Of course. Any time you have to do these problems that the computer does for you, you have to be able to interpret what the computer does. Who cares that the computer can run it unless you learn something from it? If you have some intuition of what the computer's really doing at a more fundamental level, then you'll get greater insight out of the problem. But if you're a young student, you've never seen how to get that intuition, you'll never get that from just running the computer. You'd have to be a really, really good, clever, observant scientist to get all that intuition. And Don Helmberger always used to tell me that a key part in the procedure we were doing was a mathematical trick. I always thought, "This is just too profound to be a mathematical trick." I eventually realized it was a profound physical principle that he was calling a trick. I think it really helps you learn a lot–I ran the computer a lot as a young person, and after I became a professor, my students ran the computer, and they were much better at running computers than I was, but they'd bring the computer in to me and show me the output, and I could fairly quickly look at it and say, "I think you made a mistake," and they were just always astonished that I could tell by looking at their output that they'd made a mistake. "How the heck does he know I've made a mistake?" I was usually right.

ZIERLER: When you were a graduate student, did the Seismo Lab feel like the center of the world? In other words, were you aware and appreciative of seismological research that was going on beyond Pasadena?

HEATON: I was relatively aware because as a graduate student, we were encouraged to attend scientific meetings, especially the Seismological Society meeting and the American Geophysical Union meetings. I had a pretty good view of what was happening in other places. In fact, I remember when I first started going to the American Geophysical Union meeting in San Francisco, this was in the 70s, there were probably 300 people attending that meeting. The first year I went there, I knew virtually no one, and I felt like a real outsider. "What is all this stuff?" But every year I went after that, I knew more people, and by the time it was the mid-80s, I thought, "Well, eventually, I'm going to know everybody at the AGU." But the AGU just kept growing. Now, the number of people that go to that meeting is, like, 15,000. Who could know 15,000 people? It's just completely overwhelming again now. And that's too bad because it was pretty important to be able to get a grasp of the entirety of what's going on. And it's just impossible now. You tend to get very specialized these days. If you worked in generalities, and I tend to work in generalities, it's almost impossible these days. And that's why I say in the early 70s, it was kind of perfect because the field was so small that it was possible for one brain to kind of take most of it in.

ZIERLER: Besides Helmberger, was anyone else on the faculty a mentor to you or really helpful in your education?

HEATON: Absolutely. Probably the other most important person was Hiroo Kanamori, who became faculty my second year at Caltech. When he first came, he was from Japan, and he had good English but not great. And I thought, "Why did Caltech hire this obscure person from Japan?" Again, I was very naive. And over the next ten years or so, I got to interact with him more and more because our interests were intersecting, and I eventually started working with him. And I began to realize, "Oh my God, this guy is just off-the-charts brilliant." He could solve any problem. If I was stuck, the first person I'd go to was Hiroo because he was just amazing at solving problems. For a while, I felt, "Wow, I bet I'm the only person on this planet that realizes how brilliant that man is." Of course, it's just like the AGU. Eventually, everybody knew Hiroo was brilliant, and there was nothing special about saying that. But for a while, I felt very special because I knew what an amazing talent he was.

ZIERLER: How did his brilliance shine out? What would make it apparent to you?

HEATON: We'd talk about a problem, and the next day, he'd come in, and he'd have written down the solution in a very thorough and thoughtful way. Just a very rapid problem-solver. Any kind of problem. He'd pull out various pieces of physics from other kinds of physics problems than seismology and use them when appropriate. He just knew how to solve physics problems. Truly brilliant man, and great sense of humor. The other person, of course, was John Hall in engineering. John's another brilliant mind, especially with mechanics. Patient and taught me virtually anything I needed to know about buildings. Early in my career, I think I mentioned Jim Knowles, who was the one who really taught me mechanics. Another brilliant man and a wonderful person. I was really blessed to have been tutored by such brilliant and really wonderful people.

ZIERLER: What about the Seismo Lab as an intellectual magnet, scholars coming to the Seismo Lab, seeing what was going on there, sharing their research? What interactions did you have on that level?

HEATON: Probably the most important paper in my career was a paper about slip pulses I wrote in 1989, and in 1987, I believe, Caltech had Professor James Rice of Harvard University come to the Seismo Lab as a visiting professor. He taught a class in fracture mechanics, basically. That class was very important to me, helped me to understand a certain new class of physics problem. It was in that class that I realized, "What he's describing doesn't match what I've been observing in the data of real earthquakes." And that's why I wrote that controversial paper that said, "Here's a different way for earthquakes to work than the prevailing theory," which at the time was fracture mechanics. Jim, another totally brilliant man, very patient, and a great human being. We're good friends to this day. It's interesting, Jim recognized the importance of the idea I was trying to put forward, but he never really worked that much on it himself because Jim liked to solve problems he could solve. He liked problems that he could write down all the equations and all the conditions, then, working as a brilliant applied mathematician, he'd find the solution to that set of things. The problem that arises when you turn things into these pulses, Jim used to call them Heaton pulses, are so foreign to traditional physics that all the traditional techniques don't work. In fact, you need a whole new class of equations to solve them. And it took me a while to really understand that they could never be solved with the standard techniques. And even today, I put all this stuff in my class notes that I'm still writing because I know if I wrote the papers on it today, the reviewers would just be very cruel and say, "No. No way."

ZIERLER: We talked about some of the things you lose when you outsource to a computer. Looking back to your work in graduate school on numerical simulations, flipping the question around, if you had access to more advanced computers, would the simulations have been better? Would they have told us something else?

HEATON: No. I don't think so. Given the tools that I had available at that time, I was not limited by the computer. The tool was the theory that was provided by Don Helmberger to work on this. Actually, Don didn't invent the theory either. It actually evolved over time from the 30s and 40s from some fairly clever mathematicians. As long as I was using that theory, the computers of the time were adequate to solve things. Faster computers would've probably confused me. It allowed me to think more physically about what I was doing. The important part of having a slower computer is, you have to really think about what you're doing before you do it because that's all you can get out of it. If you're limited by the size of your computer, you better use your brain better about what to do with it. But a lot of great science was done with slide rules and things like that. They didn't have a computer at all. A computer helps, but it can just confuse you if you're not careful.

ZIERLER: Who was on your thesis committee, besides Helmberger?

HEATON: Paul Jennings, David Harkrider, maybe Clarence Allen, I think. The surprising one of all that was Paul Jennings who was in the civil engineering department. Paul Jennings actually did read the theory part of the thesis. I don't think anybody else did. Helmberger thought he knew it, same for Harkrider. Don Anderson might've been on the committee. I think he probably was. But Jennings was the only one who really caught any technical errors I made in the theory part, which really impressed me because he was the one person who didn't really have an interest in that part of seismology. At least I thought he didn't, but he actually read the thesis. I'm sure it helped me later because Caltech eventually hired me as faculty, and Paul could've blocked that position easily. I think the fact that he was so familiar with my PhD work helped me.

ZIERLER: Anything memorable from the oral defense?

HEATON: The PhD defense was fairly simple. The one that was memorable and difficult was my qualifier exam. As I said, that was tough on everyone, to be presenting these half-baked ideas in front of a bunch of faculty. I remember Barclay Kamb, Don Anderson, Tom Ahrens were all on that defense. Barclay Kamb was also a really brilliant guy who asked the hardest questions, always. And he always asked questions he didn't know the answer to, of course. We always said having Barclay on your committee's like getting the "Kamb shaft." When I left the committee room, Don Anderson had a little Styrofoam cup he'd been doodling on and playing with, and I picked it up, and it had a picture of a wood screw on it. But somehow, I did okay. The last person on that one was Bruce Murray. He was a planetary scientist who eventually became director of JPL. Bruce Murray taught a class in speech and presentation, which we were required to take. I remember I was very afraid of public speaking at that point in my life. I got a C in that class. I wasn't very happy about that. And then, later, when I went through my graduation work, when they knighted us or whatever, the speaker for the commencement was Bruce Murray because he was director of JPL. And his speech was pretty mediocre. And I wanted to jump up and say, "You get a C, Bruce, you get a C!"

ZIERLER: Did you have your next opportunity wrapped up even before you defended?

HEATON: Yes, I did. I had offers from both the US Geological Survey and from a private engineering consulting company called Dames & Moore, and they worked on nuclear power plants, refineries, and things like that around the world. I had done some consulting for them as a graduate student. The problems were interesting, and the money was really useful because as I said, I had three children by that time. I first took a job with Dames & Moore which promised me they wanted to have a true intellectual in their employ. They said I could work half time on anything I wanted to and be at Caltech, and the other half had to be chargeable time at Dames & Moore. Eventually, if I'd have stayed, it would've probably led to being a principal in the company. But the half time I was with them, I was over at Westwood, where their office was, and the commute was horrible.

And in that business, you have to write down every 15 minutes of your time so they can charge it. Accounting for your life in 15-minute segments is not much fun. I quickly discovered that you could be a good scientist, but you can't do creativity that way. There's no way that you can charge a client to work on something new and innovative. You've just got to reproduce what you've already done. Because you can't predict how much time it's going to take to do something new and innovative, and you can't predict how successful it'll be. I very quickly learned that that really wasn't my future because I really like creativity. Within half a year, I went back to the USGS and said, "You guys still interested?" And they were at that time. There were two different groups. There was an earthquake prediction group and an earthquake hazard group. In some ways, I was probably more natural to the earthquake hazard group, but the earthquake prediction group told me, "If we hire you, you can stay in the Pasadena office," which was on the Caltech campus, so that's what I did. My kids were already established in Pasadena.

ZIERLER: Now that we've covered up to Caltech and your first venture beyond Caltech, one last question for today's session, since we're thinking about your time as a graduate student at the Seismo Lab. What did you learn in terms of your approach to the science that stayed with you your whole career, even when you were a graduate student?

HEATON: Well, I would say the most important thing is to ask good questions. I think people often look at some data that's complicated, and they tend to think, "Just explain the majority of the data, and forget about the outliers. There's probably something wrong." I learned that if there's some that doesn't look right, just keep digging at that thing until you understand why it doesn't look right. that may be the most important piece of data in the entire thing. Outliers are really often the key to important problems. I think that's a really fundamental thing. I call it the Sherlock Holmes effect. Sherlock Holmes would walk into a problem with all kinds of confusing data, and he'd walk over, and find the key piece of evidence, and say, "This key piece of evidence is critical to this entire case." I think that's the real job of a scientist, being like Sherlock Holmes.

ZIERLER: That's a great place to pick up for next time, when we'll get into the start of your career at the Survey.

[End of Recording]

ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Thursday, April 28, 2022. I'm delighted to be back with Professor Tom Heaton. Tom, it's great to be with you again. Thank you for joining me.

HEATON: Thank you for interviewing me.

ZIERLER: We're going to go back to 1979. To set the context for what you did right after, first with Dames & Moore, and then at the USGS, in graduate school, did you assume or were you on a trajectory that would've put you on the professorial track, and then things changed? What happened?

HEATON: I think when I was a graduate student, for practical reasons, I had gotten involved in doing some consulting work. Originally, when I was a grad student in the Seismo Lab, my second year, the Seismo Lab had a huge clump of students, and they didn't really have the funds to support all the students, so we got a message, and it said, "Please consider getting some alternative source of income for the second summer." I was the only one in the group who took that seriously, and I went out and got a job with another engineering company, Converse, Davis, Dixon, and Associates. It taught me a whole bunch of new skills, but more importantly, it gave me a new set of connections with the people doing the work to design important facilities in Southern California. Then, I continued my basic research, but I still had connections into that community of people, which is how Dames & Moore started asking me to work on some consulting things. And they knew I was interested in earthquakes and the shaking in earthquakes, so I was kind of an up-and-coming talent at that point, and not many other people were even looking at that problem.

It's how I got this job offer from Dames & Moore when I got my PhD. I think I explained that at the time, I had three children, and all my reserves were gone. I had to graduate. And I got job offers from a number of places, including the USGS, which was sort of appealing, but I was already working with Dames & Moore, and they made me a very nice offer. I took it, and it taught me something important, which is that people in that business don't really want to pay you to do new things, they want you to keep doing what you've already been doing, which wasn't appealing. But at that time, I didn't really understand, not that I actually understand today either, but sometimes you just have to go through the door and experience something to know what you're really looking for. I knew I wanted to use my science for something that affected society. That's always been appealing to me. The USGS, it turned out, was a very good way to do that. The USGS was the consultant to the United States, saying, "Here's our professional opinion on important problems like hazards in different parts of the Western US." When I joined the Survey, people were trying to predict earthquakes, so that my first job was predicting earthquakes. I did my job, but I quickly learned that whatever things we were doing to try to predict earthquakes just didn't seem to make scientific sense.

There was no way to make the logic close in a way that looked like it would lead to a system that would actually reliably work. And I still am convinced that earthquakes are fundamentally unpredictable. It did teach me that. Then, when I was in the Survey, and I discovered that earthquake prediction was probably going nowhere, to me, the more important problems were the kinds of motions that happen in cities and if the buildings we were building would survive those kinds of motions. At that time, I didn't really have much access to the building part of the problem. I mainly just worked on the ground-motion problems in the 80s. But then, when I met John Hall here at Caltech, he was a young professor in civil engineering. John was very smart, had very quick understanding of how things worked in the structural engineering world, and he was very patient. He was patient enough to actually collaborate with me, even though he knew far more about what he knew than I knew about that subject.

But of course, I knew a lot about earthquakes that he didn't really understand either. When we started talking to each other, we described what we were doing and why, and we pretty quickly realized that the two problems were not meshing in any way, that the engineering community was making assumptions that were inconsistent with at least what I thought about earthquakes. And I learned quickly the same thing, that it just wasn't connecting up. And because of that, I think that's part of the reason I got a faculty offer at Caltech. I was told later that these discussions are all secret, but later, colleagues told me the position I got was greatly debated as to whether or not it made sense to have a joint position between a person in engineering and seismology. I'm sure it would depend on who you talked to about whether or not they think it was a success.

ZIERLER: To go back to your first work in private industry at Dames & Moore, were you ever concerned because you are, of course, academically inclined, that you were leaving the academy, and you might not have a way back in? Did you see that as a risk?

HEATON: At the time, the money sounded good. They told me, "We want you to stay in academia," so they gave me a position where I only had to work half time on these chargeable problems, and the other half of the time I could spend pursuing research. And that sounded good. But when I really got into it and tried it, I discovered, which I should've known already, that to really do good research just consumes you. To get somewhere in research, you have to load your brain full of all the parts of the problem you're working on, and then the problem is solved at night while you're asleep. Then, you wake up in the morning, and it typically comes out in the shower. "Oh, that's how it works." And if you fill your brain with all the other details of some chargeable job, it disrupts that process of background processing that allows you to actually make the important realizations. People have this view of science that there's this scientific method, where you follow this set of rules, and you get to science. Of course, everybody at Caltech knows that's not how it's done. Somebody comes up with a brilliant hypothesis, and then everybody attacks the hypothesis. If it survives, it's famous. But you have to make the hypothesis, which is a jump. You put this stuff in, and, "I don't know what the answer is," but you won't find it by just standard processes. It has to be an induction. And your brain has to be ready to do that induction. And it requires all of your attention. It's hard on families because dads or moms are often thinking about something else while the rest of the family's talking about school or something like that.

ZIERLER: Tell me about Dames & Moore. What did they do, and where did you slot in?

HEATON: Trent Dames was, I think, a Caltech graduate, and it was a private consulting corporation that worked on especially geotechnical problems for the foundations and design of large structures and manufacturing facilities. They'd been around since the 30s and obviously started out as a tiny thing. It was what they call a limited partnership. If you were successful in that group, eventually you became a partner. Most of the actual work and reports were written by people who did the work, and the partners kind of talked about things, oversaw stuff, and set up the deals. When I was there, Dames & Moore had maybe four offices around the globe, and they had a big office in San Francisco, one in Westwood, I think they had one in Zurich, and one in the Midwest. If I'd have stayed with Dames & Moore, I would've eventually become a partner, probably. And it was exciting because you had to fly to all these places.

I was working a lot on a project for Exxon. That's how I got involved in the subduction earthquakes in the Pacific Northwest. Exxon was thinking there might be oil off the coast of Washington and Oregon, so they were trying to understand what the design issues would be for offshore oil platforms, and they wanted me to work on what the ground motions would be like in the offshore Washington and Oregon environment. And virtually nobody was working on that. It was ten year after plate tectonics was really accepted, and people had just realized that Washington and Oregon were part of the Cascadia Subduction Zone, which is where most of the world's large earthquakes have happened in subduction zones. But in the Pacific Northwest, there hadn't been any subduction-zone earthquakes. It was just assumed that part of the world didn't have subduction earthquakes because there were none in its history, but it turns out its history is very short. I started working on that with the assumption that maybe they could happen. And there was a person at the Nuclear Regulatory Commission who saw the work that I was doing, and later, when there was an application from the WPPSS Public Power Supply Nuclear Station in Oregon, the Nuclear Regulatory Commission knew I had worked on similar problems, and more than that, they knew I was kind of a bold thinker in the business, willing to challenge the status quo.

They asked the USGS to get me involved in the review of that nuclear power station because they wanted a more critical view than was being given by the applicants. If I'd have stayed with Dames & Moore, I probably would've been one of the people writing for the applicant. I don't know how I would've handled that. It's interesting, when I was at Dames & Moore, there was a proposed liquid petroleum gas plant being proposed for Hollister Ranch in Santa Barbara. It's this really pristine part of the coast near Point Conception, and liquified petroleum gas is great unless it ever escapes because it's so volatile. You don't want to put it in a city because it could make a horrendous explosion. Anyway, I worked on that project, and one of the people arguing against that project, those people were called interveners, and one of them was Professor James Brune, who was kind of a well-known seismologist at UC San Diego. The Dames & Moore people wanted me to dissect his arguments and take them apart. And I remember, I read them, and there were some things he did that were not correct, but I told my bosses, "At the end of the day, I think he's basically correct about these things. I think the numbers he's come up with are probably about the right numbers." To me, that was another important point, where the people who were paying for this didn't want to hear that. But from my perspective, what's the point of having all this education if you don't tell people the right stuff? That was another important time when I thought, "Maybe I'm in the wrong business."

ZIERLER: What know-how was relevant from the Seismo Lab as you went into private industry?

HEATON: I understood about the sizes of earthquakes, how slip on faults affected the nature of ground shaking. People were ultimately interested in the motion of the ground in some future earthquake, and my PhD was all about how you'd go from describing what's on the fault into projecting what the ground motion would be like. Not many people were doing that problem at that point in time, and there just weren't the techniques to do it. Don Helmberger, really, at that instant in time, made that possible. Today, we use very different techniques than what were used then, but at that time, I was one of the few people who knew how to do that problem.

ZIERLER: And how did the opportunity at USGS come about, and why was it attractive to you at that point?

HEATON: As I said, about six months into my job with Dames & Moore, I decided maybe I'd chosen the wrong direction, so I went to my colleagues at the USGS, who was hiring at the time because the National Earthquake Hazards Reduction Program, which had just been funded a couple years earlier. There was a lot of funding in the earthquake business, and they were hiring people, and there just weren't that many of us who were qualified for what they were looking for. When I told people in the Survey that I was interested in making a change, there were two groups in the USGS that responded. One was the earthquake prediction program, and the other was the earthquake hazard group. In some ways, my background was probably more amenable to the earthquake hazards group, but the earthquake prediction group–the guy who made me the offer was Dave Hill, who was a Seismo Lab graduate, and I knew him very well and respected him. He told me, "We'll just station you at Caltech," so I didn't have to do anything except change who I was working for, but I could work on many of the same problems and get paid differently. I took a big pay cut to do it, but my children survived it.

ZIERLER: To what degree was USGS on your radar when you were at Caltech? Was it an esteemed institution? Was there collaboration?

HEATON: The USGS is an esteemed organization in the field of geology. It was the first science agency in the United States, and they were tasked in the 1860s with quantifying the geology of this new continent with a lot of emphasis on mineral management types of things, where the resources are and how to deal with them. Then, in the earthquake business, the USGS wasn't so expert. But along came the 1964 Alaskan earthquake, and that was an enormous earthquake that had a tremendous impact on the state of Alaska. And as a result of that earthquake, there was a new program that was put together, the National Earthquake Hazard Reduction Program, and that basically took all work in the federal government and put the science work to be managed under the US Geological Survey.

Other people who were in other federal agencies working on earthquakes were transferred into the USGS in the 1970s, and it was a little bit of a chaotic situation because suddenly, this group of people that hadn't been working together was working together, and they weren't even quite sure what it was they were supposed to be doing. It had not really yet congealed into its destiny in the geophysics work. The geology part of the USGS, mapping and all that, was very well-established and well-respected. But geophysics was just kind of forming, so I got to get in at the ground level. And it turned out at the beginning, there was plenty of funding, but we didn't have any earthquakes for a long time. Then, the US started in the political wars. We were level-funded for 20 years, and there was a lot of inflation in there, so by the time I left the survey, funding was just really tough to find in the USGS. I went in at a period when it was flush, and when I left, it was desperate for funding.

ZIERLER: What year did you start at the Survey?

HEATON: I started in 1979. I actually defended my thesis in 1978, and then I started in the Survey one year later.

ZIERLER: You were only with Dames & Moore for a short period of time.

HEATON: For one year. Plus, I was a consultant before. I left many friends there. Even after I left Dames & Moore and I was working with the USGS, I still had contacts with the people in Dames & Moore. There are still people I communicate with there who are important to me professionally.

ZIERLER: Going back to an earlier question, when you got to the Survey, did you feel like that was a step back towards academic life, to some degree?

HEATON: It clearly was. Nobody was really telling me exactly what to do. The Pasadena office I was in was a sub-office of the Western Region office in Menlo Park, California, so my orders came very circuitously down to me. By the time they got to me, it was, "Do something good." [Laugh] I had some general guidelines, but generally, I was able to pursue my own interests. And that worked out fine because my interests, I think, were compatible with the overall interests of the USGS.

ZIERLER: On that point, what were those overall interests at the USGS, and how did you fit into that?

HEATON: The USGS is there to help guide federal policy with information from scientists, and the view was always, "Don't do the politics, do the science. Always think about the importance of the science to the national endeavor. Keep your nose out of politics, but make sure that what you're working on is relevant to societal good." And that worked well. I remember when I was working on the Pacific Northwest, I wrote some provocative paper saying, "This place looks an awful lot like the other parts of the world that have really big earthquakes. It might be they just haven't waited long enough." The chief geologist of the state of Oregon wrote a letter to the head of the USGS at the time, who was another Caltech graduate, Dallas Peck, and he said, "Can't you get this guy Heaton to work on something useful?" I still have that letter. It's kind of a strange letter. I remember being a little concerned about having the chief geologist of a state write such a letter to the head of the USGS. Word came back from the USGS management and said, "We got the letter. We like your work. Just do good work, don't do mistakes." People were noticing this work, but they didn't tell me to change it. It's interesting, after I left the Survey, things have gotten much more controlled in the USGS. Now, if you write something with a USGS coauthor, it has to be approved all the way up in the Department of Interior. It can't look as if it's got anything that might be politically sensitive in it. It just takes forever. There was a lot less–I hate to use the word censorship, but it's almost like there's censorship now. People are really far more restrictive about what they can say in the USGS.

ZIERLER: As is often the case in government facilities, there's access to cutting-edge instrumentation. I wonder if that was your experience at the Survey.

HEATON: Certainly, the USGS had the resources to run the seismic networks. The chief instrumentation for me has always been seismic networks. The earthquakes are the experiment. Some people try to create earthquakes in the laboratory. I have become convinced that that's kind of a hopeless procedure because earthquakes in the laboratory are so much smaller than the real thing, and to think that the physics doesn't change when you go from this tiny thing up to the large thing, I think, is completely crazy. Many people act like seeing something at the inch scale is the same as seeing something at the 10-mile scale. I'm quite convinced that's led us in very crazy directions. I don't use a laboratory. My laboratory is observing earthquakes. The USGS had the resources to run the seismic networks, and prior to the USGS earthquake program, the network in California was run by two agencies. One was UC Berkeley. That was funded by the UC system after the San Francisco earthquake.

And then, in Southern California, the network here was funded by the Carnegie Institute and Caltech. By the time I was a graduate student, the Caltech network was mainly funded by the US Air Force, believe it or not, because the Air Force was interested in detection of nuclear explosions and tests. The Seismo Lab had a huge blanket project with the Air Force, and basically all the activities of the Seismo Lab were covered under that Air Force contract. And they didn't really have anybody in the Air Force who knew enough to really intervene and tell the Seismo Lab what to do, which was pretty nice. But eventually, like I said, after the Alaskan earthquake, the USGS became the federal agency that ran the seismic program, and they installed the networks, and they came to Southern California and installed a network in the 1970s that was based on 1970s technology. It was designed to pick up the epicenters of all the tiny earthquakes to look for patterns that might proceed some future large earthquake. That turned out to be a dead end. But the Seismic Network had 250 stations. It was an enormous endeavor, and it wasn't especially useful to me because I was interested in understanding the actual motion of the ground and how it related to the faulting in the earthquake.

The network the Survey had built went off-scale for any kind of important earthquake. It wasn't useful to me. When I joined the Pasadena office, I was in charge of that network, and I very quickly started to re-channel the direction and design of that network to better be able to understand what was happening in the earthquakes. And early in the entire endeavor, I became excited about earthquake-alerting. I knew we couldn't predict them, but I knew we could tell people that waves were on the way. Earthquake early warning. From the very beginning, I was trying to figure out how to evolve our existing network into one that could do early warning. Because funding was so level, I couldn't get the funding to build it, but I kind of designed in my own mind what we needed to do. After the Northridge earthquake, suddenly, new funds became available.

The first set of new funds was before Northridge, and those came from the Whittier Foundation. Caltech found those funds, and Hiroo Kanamori and I collaborated on where the network should go. He had 16 of those stations put out following the Whittier Narrows earthquake. The Whittier Foundation was a private foundation, and they funded this new broadband network in the late 80s, and the impetus came after the Whittier Narrows earthquake. After Northridge, some real money became available to build a new network of 150 of these stations. And Egill Hauksson was hired in that period of time, and Egill really oversaw the implementation of that new network, and it just completely changed the way we did seismology in California, and ultimately, across the globe because now that's the standard for the way earthquake networks work.

ZIERLER: On the one hand, you're convinced already at this point that earthquake prediction is not going to happen, yet there is advance in early warning. What's exactly the difference between the two?

HEATON: Earthquake prediction is the idea that I could tell you before the earthquake even starts that there will be an earthquake in someplace and how big it will be approximately, and that you would have tens of minutes, to hours, to days to prepare for it. If you actually could do this, you wouldn't have any more earthquake deaths because you could be out in the street. You don't have to evacuate anything, you just have to be out of harm's way, and you could turn off all the flammable processes like refineries, put them in a safe mode where they wouldn't explode, get people off the freeways. You could save a lot of lives. But we don't know how to do it. But it turns out some of these things I just talked about, preparing people on the freeway, or extinguishing flames, or getting out of the way of things that might fall on you, they don't actually take very long to do. Maybe you could wait until an earthquake starts, and then tell people, "It's time to do that thing right now, but you better do it fast because you've only got seconds to tens of seconds before the shaking hits you."

The first examples came from Japan, where in the 1960s, when they decided to build high-speed trains, some people said, "Wow, we have earthquakes here. You really want to be on a train doing 150 miles an hour during an earthquake?" As part of their design of the new train system, the Shinkansen–shin means new, kansen means new trunk railway system–they decided to build in a system that would control the trains as efficiently as possible in an earthquake. Their first system was just an operator shutting off power to the tracks, and the operator was watching a red light that was triggered by a seismometer along the tracks. If he saw some area of the tracks shaking, he'd turn off the electricity. Then, the Japanese went, "That's kind of crazy to have a guy doing that. If it turns on a light, can't we just automatically turn off the electricity?" They automated it.

Then, they thought, "Well, most of these earthquakes are not along the tracks. Why don't we let it turn off the electricity from seismometers that are in other places near the railway system, so it'll be faster?" They kind of evolved their way into this early warning. When I was at Dames & Moore, some of the engineers were telling me about what the Japanese were doing, and I thought, "Wow, that's really cool. We could take that a lot further than what they've done with it by automating it through a computer network." That's when I wrote a paper about if we computerized it, what its attributes would be. What I wrote in that paper is basically what we eventually built.

Earthquake early warning, you don't actually predict the earthquake, you just predict the aftermath of an ongoing earthquake. You do it so fast, you beat the waves before they get to you. But the best thing to do is, if you've built your society really robustly, then you don't really have to rely on whether or not you've got an early warning system or an earthquake prediction system. If you can be confident that an earthquake is not going to injure you, then they're just entertainment. People used to pay money to go on thrill rides, and one of the thrill rides, they put you on a table and shook you around like an earthquake. People would pay for that because they knew they weren't going to be injured. If you could build your society with enough robustness in it that you knew, "I'm not going to die in an earthquake," what's the big threat?

ZIERLER: What were your key responsibilities when you became scientist in charge of the Pasadena field office in 1985?

HEATON: The big things were funding other research groups in Southern California. We had a staff of about 15 people, and I was the supervisor of those 15 people. And whenever there was some discussion in the USGS about strategies for what to do nationally, an important part was Southern California. And I always represented the interests of Southern California. There was kind of a territorial fight that was going on at that time because the main office of the USGS for the earthquake program was the Western Region Office in Menlo Park, California, and it had about 250 employees working on the earthquake problem, and Pasadena had 15 or so. There was a really big difference in available resources, whereas the population of Southern California is much bigger than the population of the Bay Area. There was this kind of fight going on about who got resources to which things. And I represented Southern California. To make things even worse, there was this really divergence in scientific philosophy about how to handle earthquakes. Northern California office was really into repeat of 1906 kind of earthquake, and that makes sense because it's the only real important earthquake that's happened historically in the Bay Area, except maybe Loma Prieta. They focused everything on what they said was the earthquake cycle. When was the last one, and are we going to be ready for the next one? They were looking for a cycle to repeat these characteristic earthquakes, they called it. And Southern California's had a very divergent history of earthquakes, lots of different kinds of earthquakes in many different faults and places, a much more complicated history. Down here, it didn't make any sense to focus on just one particular earthquake. In fact, the whole idea of a cycle didn't make any real sense either. The way we kind of approached the problem was quite different in Southern California than Northern California. There's still some divergence between the two now.

ZIERLER: While you were at the Survey, was there opportunity to pursue your interests in architecture and engineering with earthquake preparedness?

HEATON: I would have to say not really. I interacted with some engineers, but not strongly. Basically, they weren't interested in what I had to say because they wanted to build buildings, and information from the seismologists kind of just complicated their lives. Anything we provided to them was not helpful for them getting done what they wanted to do. Especially architects don't want to know that stuff. They want to make what they want to make. If it's okay with the seismologists, that's good. If it's not, "Why do I want to talk to a seismologist?" They don't. It was only when I joined Caltech that really came up. There's this idea that I would be in both departments, and in some ways, it makes sense, but in reality, it's pretty tough. And still, in the real engineering world, real engineers want to build things. And to have people like me out there saying, "You're probably going about this the wrong way," they just don't want to hear it. Unless I go out there and build it myself, which is not going to happen. They don't want to hear that. It's just bad news for them.

ZIERLER: Tell me about the 1989 Loma Prieta earthquake. Where were you, and what was it like?

HEATON: I was on the second floor of South Mudd next to the seismometers, about to go home and watch the World Series, like everybody else. And I saw the seismometer needle just going bang, bang, bang, bang. The alarms were going off, and I knew, "I guess I can't go to watch the World Series." I knew it wasn't here because we weren't feeling anything, but I knew something really important had happened. We worked hard to try to provide the information that we could to our colleagues in the Bay Area. There's kind of an unwritten rule in the Survey that if an earthquake is in your area, it's your responsibility to coordinate with local agencies, provide information and all that. We tried to provide what information we could to our scientific colleagues in the Bay Area, but they were up to their necks in data and requests, and there were ten times as many of them as there were us. I could mainly just watch from afar. There was some issue where there was an earthquake about six months ahead of time in the same place as the Loma Prieta earthquake, and the Survey was saying that the Loma Prieta earthquake was on the San Andreas Fault, and that it was more or less predicted. You'd have to have read what they'd done beforehand to come to that conclusion.

You'd have to read it very carefully. There's no way, in my opinion, that they really had predicted that earthquake. Furthermore, it became pretty obvious that it was on an associated fault to the San Andreas Fault. It wasn't actually on the San Andreas Fault. At that time, I was on something called the National Earthquake Prediction Evaluation Council, which was a group that was supposed to provide advice to presidents and governors in the event somebody made a prediction of an earthquake. I was also on the California Earthquake Prediction Evaluation Council. We had to deal with that earthquake in both of those councils. I got quite familiar with it. Then, Dave Wald, one of the graduate students here at Caltech who did most of his PhD work with me when I was at USGS, he and I wrote a bunch of papers about the Loma Prieta earthquake explaining what happened on the fault and why the ground shaking was the way it was. It was a reasonably well-recorded earthquake at that time.

ZIERLER: Once you had a chance to look at all the data from the earthquake, what were some of the big takeaways? What opportunities did you have to demonstrate what had happened?

HEATON: Like I said, there were some pretty good recordings of the ground motion in close, and it was clear that, one, it wasn't on the San Andreas Fault, it was on a subsidiary thrust fault in the area that was quite deep. And it was sort of a double earthquake, there was a section that went south from the hypocenter and another section that went north. And there was some very violent shaking in the Santa Cruz Mountains in that earthquake. That area's got a bunch of Redwood trees in it, and above the earthquake, virtually every Redwood tree had its top 30 feet snapped off. The ground was littered with the tops of these trees. It takes pretty violent ground-shaking to do that. You don't really want your building in that. And there are a few houses up there. Turns out houses are very resilient in earthquakes. The people who were up in the area had amazing stories to tell about pianos being flipped upside down in the shaking. It was obviously a very violent shaking. It was pretty clear that there were some localized areas that got this super-intense shaking. Fortunately, they were extremely sparsely populated, and the structures that were there were these wood structures that are very robust. But if you took that same kind of motion and put it in a city, oh my God, it'd look like Mariupol or whatever. It'd be really bad. Part of what we did was try to make people understand why there were these intense areas of high shaking and what those areas of intense shaking looked like. That was an important part of the Loma Prieta effort.

ZIERLER: Before we get to your tenure as project chief, some general questions. The Southern California Seismic Network, first, what is its mission, and how far back does it go?

HEATON: The Southern California Seismic Network, I'm not sure how long we called it that, but the original network was conceived by Harry Wood, who was one of the first scientists brought in by the Carnegie Institute. Harry actually argued that the seismological network and laboratory should be set up in parallel to the Seismographic Station in the Bay Area. He made that argument in 1916, and it became a reality in about 1923, I believe, which is when the old Seismo Lab was first established in the San Rafael Hills and operated by the Carnegie Institute that ran Mount Wilson. Then, Wood collaborated with John Anderson, who was an astronomer on Santa Barbara Street working on the Mount Wilson Telescope, and he built telescopic instruments. Anderson worked with Wood to make a new generation of seismometer that turned out to be a very effective, simple, robust design, and there were about a dozen of these stations put out around California. Those stations were used to characterize all the local seismicity, and the records from that system were used by Charles Richter to develop the magnitude scale.

That was the original Seismic Network in Southern California. I'm not sure what they called it at the time. It might've been just Caltech's network, I don't know. And then, in the 40s, 50s, and 60s, but especially 60s, there were some new instruments that were put out. The director by that time was Frank Press. He used to be the National Science Advisor and a famous scientist. Frank Press worked more on global seismology. He worked on earthquakes around the globe, and he really changed the Seismo Lab. Prior to Press's time, the Seismological Laboratory was kind of a standalone research organization for earthquakes operated by Caltech, kind of like Palomar or something like that. And then, when Press got involved, he said, "This is an academic enterprise." He really pushed that there should be more classes and many more students.

That was the time when lots of the famous geophysics grad students Caltech had and sent off to other places were under Frank Press's time. Prior to that, I think Richter had one student in his entire career, and Benioff swore he'd never work with a student. Gutenberg didn't have many students either. There just weren't students in the Seismo Lab. But Press brought in lots of students. And that turned out to have really changed the focus of the Seismo Lab forever. Today, in the Seismo Lab, we still run a network, but most of the endeavor is about basic research done mostly by graduate students with professors on a variety of problems. Plate tectonics, mineral physics, earthquake physics, and the like. But it's not nearly as focused as it was prior to Frank Press's time.

ZIERLER: You mentioned the increasing budgetary difficulties at the Survey. What were the funding sources for the Southern California Seismic Network, and how did that influence your day-to-day as project chief?

HEATON: The fundamental piece of equipment for all of the USGS was the Seismic Network. If we didn't have a seismic network, we couldn't do any of our research. We had to run a seismic network, and even when we had budgetary problems, we always had to somehow keep the Seismic Network going. But we didn't have the money to develop it and make it into something new. Not being able to get the money to make new developments was pretty frustrating, especially when we knew that there were some important developments that would very much change the field if we had the resources to do it. And again, the USGS earthquake program, the entire thing was struggling. We had to keep the networks going. But many other people had projects they thought were important, so we were trying to compete in the system with other projects they thought were more important. That was the source of a lot of tension.

ZIERLER: What was the reporting structure when you were project chief of the Network? Who reported to you, and who did you report to?

HEATON: We had a series of annual reports that were published in the catalogs. They'd be maybe 300- or 400-page books of reports from different projects funded in the earthquake program. We had to write one for the Network, and then there were ones for people doing geology, ones for people looking at source physics. I had to write one for working on Early Warning and things like that. We had to write reports very much like NSF required reports, although the USGS reporting requirements were not nearly as onerous as what reporting requirements are today. We basically all had to send in 25-page reports every year. But that's not too bad.

ZIERLER: What do you see as some of your key achievements in the role of project chief?

HEATON: I think the most important thing was to focus the evolution of the Seismic Network toward a much broader role of reporting true ground motion. When I started, the seismograms basically were turned up to maximum gain and put onto analog tapes or even pieces of paper. As soon as an earthquake would start, all you could see would be the start time when the wave arrived. But all the rest of the signal was just bashing back and forth between the limits. You couldn't actually figure out what the real motion of the ground was. And I thought it was critical that whatever we recorded should actually tell us what the real motion of the ground was because that's how we really understand the processes, how the waves propagate and how the source worked in the first place. I pushed hard to try to calibrate the system and to get new types of sensors in the system so that we'd know what the actual ground motion was. It's interesting, when we finally pulled it off, I always told people, seismologists in the future will all worship at our graves for making this important change that allows them to use this data. But as it turns out, now that I'm older and there are new seismologists, they just take the system for granted and assume it's always been like that. They don't even have any understanding of how the old system worked. They've never seen it, so to them, it's just always existed. But of course, it had to be created.

ZIERLER: In getting that system up and running, who were some of the key institutional partners for you?

HEATON: The biggest collaboration was between the USGS Pasadena office, and especially Caltech, the Seismo Lab. The key person at Caltech was, for sure, Hiroo Kanamori. But there's no way that Hiroo and I could've done that by ourselves. There were a lot of other people involved. Rob Clayton, Lucy Jones, especially Egill Hauksson was involved in it. Kate Hutton was involved. Many of the standard seismology people were involved between Caltech and the USGS. We had to coordinate with the state of California, who was running their own network of just strong-motion instruments, and there was a lot of turf involved in that. I don't really want to go into the details of that at the moment. Ultimately, we collaborated, but that was kind of tricky. You mentioned SCEC. I haven't talked much about SCEC. As I said, we were struggling with budgets in the USGS earthquake program, and part of the program was that our budget went out for research to be done by a number of universities. Caltech was one, UCLA, UC Riverside, places like that were all doing research and getting some of that money. In 1987, we had the Whittier Narrows earthquake, and it had a lot of political attention. The politicians said, "We need to react and help the earthquake program." They had some hearings at Caltech about where earthquake research should go in California in 1988 or so. They called in the politicians, set up a half-day symposium for the scientists to come in and say, "What should we do?" They invited famous scientists from different universities.

And each group came in and said something different. The politicians were just really angry. They were livid. They said, "We threw you a softball, and you completely whiffed. You should've told us, 'Do this one thing.'" But they were told all these different competing things. We decided we needed to speak with one voice and get our act together. That's when we decided we needed to have an earthquake center that allowed us to focus our activities among many different organizations. And at that time, NSF was starting these Science and Technology Centers. Caltech was intending to send in a proposal for a Science and Technology Center for earthquakes, but NSF said there couldn't be any more than three Science and Technology Centers per institute. And the management of Caltech had already decided physics, I think it was LIGO, and a couple other centers, took precedence, and that Caltech was not going to send in an earthquake center proposal because it wouldn't be one of the three high priorities. But USC, at the time, saw it as a golden opportunity for them to kind of insert themselves into this problem much more than they had been. USC was chosen as the organizing institution for SCEC, and I spent a fair amount of my time helping to put together the SCEC proposal and get it funded.

ZIERLER: Do you think it was the right move for USC to host SCEC?

HEATON: I think it certainly helped their program tremendously. USC made some very generous decisions about not taking overhead and not over-promoting their role in SCEC to begin with. Of course, that was 30 years ago, and over time, USC has become ever more important in SCEC. It's natural because they're the host institution. That meant there's now another place out there working on earthquakes. There's USGS, the traditional Caltech, SCEC, and Caltech's part of SCEC. And SCEC has gotten their own PR part of things. Caltech's central role it had throughout much of the 20th century has disappeared mostly.

ZIERLER: Moving into the mid-1990s, what were the original conversations that ultimately led to your faculty appointment at Caltech? Who was talking with whom?

HEATON: Conversations about faculty things at Caltech are usually extremely confidential.

ZIERLER: Who reached out to you, or who were you talking with that made this even come up on your radar?

HEATON: In the 1990s, I started teaching engineering seismology. I think three years before I joined the faculty, I started teaching a class, and Caltech paid me as a lecturer. It was a class in engineering seismology. It was really focused on seismology, but I was starting to add more engineering as time went on. And I think the fourth year I was doing it, Caltech had made me, prior to that, something called a visiting associate. And there are a bunch of people at Caltech who have that title. It gives you status to be on campus and be from somewhere else. But they offered me something else called faculty associate. I don't think it exists anymore, but at that time, it had to be approved by the faculty committee, the full faculty meeting. For some reason, I was told it was a more elevated position. There were only a couple people like that at Caltech. And I never did quite figure out what it meant that was different from what I was doing. But it put me more in people's view. By the time I got the faculty offer, I had been through this process of faculty associate, and I think if you wanted to know something about how the offer actually got made, you should talk to Ed Stolper. I think Ed was the key person. He's got a memory like an elephant – he never forgets anything.

ZIERLER: And what was the research and the teaching that you were doing up to that point that made this an easy transition? In other words, were you academically connected at all? Or was it really a big transition in 1995?

HEATON: I'd been teaching this class, so that made it easier because I had a bunch of class notes I'd been using. But then, in '95, when I took the position, I was suddenly also in the engineering department, and originally, I thought, "That's not going to change my life much. It's mainly what I've been doing." And the engineering part was what's sometimes called a courtesy appointment. But the engineering faculty at Caltech in civil engineering, which once had a very large department, nine people, but had been decreasing through the 80s and into the 90s. By the time this offer was made to me in civil engineering, I was going to be one of four civil engineers, and it was a half-time position. Some of the engineers came to me and said, "Getting into a civil engineering position at Caltech is a pretty weird thing," because they'd tried to build up the department before, but it was always turned down.

Along came their opportunity, so they took it, and they told me, "We don't expect you to just disappear and spend all your time in the Seismo Lab because this is important for us. It's one of the few positions we've got. We expect you to show up and participate in the engineering part." I decided there was no way I could really work full-time at two different offices, so I put most of my books and secretarial staff in engineering, so I'd spend enough time there. It turned out, I spent most of my time there, and eventually, most of my students came out of engineering. A lot of my mental resources were now associated with engineering. I think some of my earth science friends were kind of, "I wish Tom would spend more time in earth science." But there's only so much time you've got.

ZIERLER: Just to clarify, the initial position was a joint appointment? That was there from the beginning?

HEATON: That was there from the beginning. And Caltech was just starting with that. Nobody quite knew what it meant. Nowadays, it's almost certain that when we do a joint appointment, we say, "It's a joint appointment, but his/her home is this place." When I did it, the expectation was it would be truly joint. My background was geophysics, but because now I was spending a lot of time in engineering, with engineering students, and most importantly, teaching engineering classes, it meant I really had to put a lot of my attention into engineering problems. Teaching engineering classes, I learned a lot, but it was a real challenge. It's one thing to take a class, but it's another thing to stand up in front of Caltech students who were pretty bright and lecture with a piece of chalk about a technical subject. You better be ready for that. You might as well be naked.

ZIERLER: The title, professor of engineering seismology, is the idea there that within that title, it would convey the fact that you had this joint appointment?


ZIERLER: Has there been since another professor of engineering seismology? Or do we see that title at other academic institutions?

HEATON: Other people call themselves engineering seismologists, but it's pretty rare. Nadia Lapusta is joint by the way mechanical engineering and seismology, and Jean-Philippe Avouac is also joint between engineering and geophysics, so there are two others. But they work more on earthquake physics kinds of problems. I'm the only one who worked on the earthquake engineering problems.

ZIERLER: At that point when you joined the faculty, what were your research goals? You had this really interesting joint appointment, you had access to both of these worlds. Let's start with your interest in the societal impact. What were you going to do that was going to help society and even save lives for the next big one?

HEATON: To have that kind of focus when I was in the Survey, I got to call all my own shots and put my interest into what ground motions are like, how to do early warning, how to run a seismic network. But a Caltech professor's kind of different. Most of my attention over the years has been with my 30-some graduate students. My main job in life as a Caltech professor is to get those graduate students launched, get them a PhD. I'm their advisor, so I have to spend time with those students. You can't just go and say, "Here's what I want to do," and make these students do that. It doesn't really work that way. They're getting their PhD, and it's their PhD, and it's a negotiation between the student and professor about what it is that's going to excite a student enough to get them to follow through on getting a PhD, which is a big effort and investment.

I spent time interacting with them in classes, then they'd come in and say, "What am I going to do for my research?" I'd always tell them, "This is going to be your PhD, so I'm not going to tell you what it's going to be. This is your life, and this is your one opportunity in life to really go out and find your own destiny. Our job is to figure out what it is you want to do. For instance, here are some problems I think are especially interesting. Are you interested in any of these problems?" Over the years, I generally ended up finding students that were interested in most of the problems that were of interest to me, but I couldn't tell you in any given year, "I'm following this particular path of direction," because it just depended on the student. I was always interested in earthquake physics, and I had a couple of students who were interested in that problem, were very talented, and really made a difference. If I'd given the same problem to some of my other students, it would've just ended their careers. It would've been all they could handle. It would've been the end. It's a very different process once you're an advisor, at least in my world.

ZIERLER: Tell me about your work with John Hall on flexible buildings.

HEATON: It started in 1994, and we were leaders of a field trip for something called the Earthquake Research Affiliates, and we were explaining about earthquakes that could happen in the LA area, what kinds of buildings were out there, and how they reacted to different kinds of earthquakes. I was doing the faults and ground motions, and John was talking about the buildings. We ended up on the top of the library tower building, and I was explaining about the earthquakes, John was explaining about the Library Tower, a tall steel building. As we were explaining, it became pretty obvious there was a disconnect between what each of us was saying, and we looked at each other like, "We've got to talk some more." We started to look at it more, and the 1994 Seismological Society Meeting was in Pasadena, and Caltech was the host organization for that. And there was a joint meeting between the seismologists and the earthquake engineers. It was serendipitous that it was the same year as the Northridge earthquake.

As part of the joint meeting, there was a day when we made a hypothetical, "What would happen if there was a magnitude-7 thrust vault, kind of like Loma Prieta, under downtown LA? What would happen to the tall buildings?" A bunch of different engineers were given ground motions that were produced by seismologists, and we said, "What would that do to the buildings?" And they were all supposed to talk about this at this joint session, which was kind of a unique day in our business. I think John was the only engineer at the place that actually took those ground motions and used them as we intended. His conclusion was that current buildings would have a real hard time with those ground motions. The other engineers, who were in the practice, said, "Oh, our buildings are fine."

That's when John and I really started to work together. Actually, we had worked together about 1980 in a completely separate problem. John was not only brilliant with buildings but with finite element methods. I had found an inconsistency in a relation that was supposedly used in geophysics that was calculated by a group up at Stanford, which required a numerical calculation. I knew it was a tough calculation, and I thought John would have the mechanical chops to be able to solve that problem. He and a graduate student worked on it as that graduate student's PhD, and he solved that problem. I used that in some of my work from then on. I got to know John at that time and got to respect what a brilliant guy he is. He really is a special talent.

ZIERLER: What was the significance of the TriNet project?

HEATON: [Laugh] That's TriNet, T-R-I-N-E-T. People have spelled that with a Y. Originally, when USGS and Caltech decided they wanted to build this new kind of network of broadband seismometers, the Whittier Foundation Network kind of thing but bigger, after Northridge, we put in a proposal to build this bigger network, and it went to FEMA. At the same time, the state of California had a program to record strong shaking in earthquakes, and that program was the California Strong Motion Instrumentation Program. And they were kind of struggling with funding, too, so they wanted to send in a proposal to build up their network as well. And FEMA said, "There's no way we're going to fund two separate networks. We need to have just one coordinated project." TriNet was born. It was a shotgun wedding of Caltech, USGS, and the California Strong Motion Instrumentation Program.

ZIERLER: Tell me about your work with Hiroo Kanamori on the Cascadia Subduction Zone and how you became a bit of a historical detective in learning about this.

HEATON: I was assigned the job of reviewing the nuclear power plant license. When I looked through the report for this license, I thought there were many unanswered questions that left me unsettled about it. I was not that expert on subduction earthquakes at that time, so I went to talk to my colleague who was the expert on subduction earthquakes, Hiroo Kanamori. He was a much younger man then. He's incredibly knowledgeable about subduction earthquakes, a very curious man, a broad thinker. The more we talked about this problem, the more we both realized it was a really important problem and that it hadn't been solved yet. And one of the answers to the problem was the possibility that there could be very large infrequent earthquakes that seemed to be consistent with the data.

We wrote a paper together on this that got a lot of notoriety at the time. Then, we tried to think how we could solve whether or not such earthquakes actually happened. I was looking through catalogues of tsunamis to see if there were any unidentified tsunamis. I thought there might've been some possibilities, but I wasn't sure what. I was giving some talks on this subject throughout the USGS and at various universities, and one of my colleagues in the USGS side who worked on coastal geology had read a book by a guy who wrote about the Pacific Northwest in 1850. And in that book, he recounts these Native American legends of unusual disturbances of the land and sea. The way it's written, it sounds like it was not their standard–it was quasi-historic. They said, "In a time long ago, but not so long as the beginning of the earth." They described something that could possibly have been a bit tsunami at Cape Flattery in Washington. I wrote this up and put it in the Seismological Society, and I talked about it at some of these meetings. At that time, another colleague, Brian Atwater in the USGS, who really worked on coastal deposits, looked at it–actually, the story's more interesting than that. Before Brian got interested, Hiroo told me, "I have some colleagues in Japan who work on coastal deposits from earthquakes.

I will invite them to come visit in the United States, and we can take a trip to see if we can find these coastal deposits from prehistoric earthquakes in the Northwest." A couple of Japanese specialists came over, and we took a two-week trip up to the Pacific Northwest looking for these deposits. We found some interesting things, but nothing that was truly convincing of an earthquake. When we were doing that, I said, "We're only looking at things that look like they went up in the earthquake. Sometimes the ground goes down in earthquakes. We could be looking for subsidences." The Japanese geologist said, "Yes, it's true, but if that happens, that means we'd have to look for buried deposits, and it's very dirty. We don't like to get dirty." They only wanted to look for things that went up. I was talking about these things to people in the USGS, and one of them was Brian Atwater. It turns out Brian Atwater's research was all about buried deposits, and he just loved to play in the mud. He'd go out with his shovels and dig through things. Shortly after I talked to him, I got a postcard from Brian, some pictures he'd taken of a place he'd visited at Cape Flattery. He said, "You see those layers? Is that what you're talking about?" [Laugh] I still have that postcard.

Brian started a career of finding these layers that he interpreted as subsidences in the ground, probably from earthquakes. And he made a story of how often they'd happened in the last 4,000 years or so. But things were all kind of not nailed down until he found a ghost forest in the border between Southern Washington and Oregon. The ghost forest was a bunch of cedar trees that had suddenly been subsided and inundated with salt water, killed all the trees. And it turned out these cedar trees had been around for 300 years. He was able to date them from a variety of techniques, but especially from tree rings. He knew the exact time of year that all of these trees had died. At that point, he started collaborating with a tsunami scientist in Japan, and they were poring through Japanese records of what are called orphan tsunamis. An orphan tsunami is one that came in and often took people by surprise because they came in without any shaking. They want were a tsunami from some distant part of the Pacific.

He found a tsunami in January 19 in 1700 in their records. It turns out 1700 was the year that these trees all were killed in the Pacific Northwest. They looked around and saw no other area around the Pacific Rim that had an earthquake that year. Only the Pacific Northwest had really unwritten records from that point in time. There were these trees that were killed then, so almost certainly, that orphan tsunami in Japan was this last Cascadia earthquake. Now, that last Cascadia earthquake is virtually an established fact. It's quite an amazing detective story that we went from, "It's aseismic," to knowing the day of the last earthquake. It's amazing. And we know something about its size because we know the size of the tsunami in Japan.

ZIERLER: I'm curious, the debates around the so-called cyclical nature or predictability of earthquakes, does it more or less parallel those same debates about tsunamis?

HEATON: In a sense, it does. Again, people have made from the beginning this view that earthquakes are the seismic cycle. It was Ried who said the earthquake strain builds up steadily over the years at a constant rate until you get up to a high enough stress, then the whole thing breaks in an earthquake. Then, the whole process just starts up again until you build up enough stress. The idea is, if you know the average time between the earthquakes, that's pretty close to the real time between the earthquakes. But any time anyone's been able to actually really date things, in some very rare instances, we have two earthquakes on the same fault that are historic. Whenever we've looked in that respect, nothing is even at all. A good example for us here is the last earthquake on the San Andreas Fault in Southern California was 1857. The previous one to that was in 1812. That was 45 years between 1812 to 1857, and we haven't had one since. Now, it's 160-some years later. 45 and 160 is not even at all. Other places in the world, we've been able to look at geologic repeat times for some big earthquakes, and they don't look like they're even at all either. Although, geologists often try to arrange the data, so they look as even as possible because then, it sounds more plausible to them. But I think the evidence we've got is that they're not regular at all. But we know plate tectonics is regular. We can measure it. It's very steady. But the earthquakes are, in my view, very irregular. One of the big questions is, why are they so irregular? What's controlling that?

ZIERLER: By the end of the 1990s, what were some of the major improvements in earthquake monitoring, and how do you see your research contributing to those improvements?

HEATON: My research, again, was especially about the seismic networks, and how to record ground motion, and how to have a good ground-motion system. The big change that happened also in the field from the 80s into the 90s was space-borne geodesy. A big part of our field is measuring the ground shaking in earthquakes. Another important part is looking at the long-term deformation of the earth over centuries. In plate tectonics, we're moving here in Caltech about two inches a year to the Northwest compared to the center of the continent. You can measure that geologically by just looking at the total offset on the fault and trying to get some ages on geologic things, or you can measure over the last century by doing really accurate geodesy. People did measure those things, but they were very hard to measure with standard surveying techniques. They used leveling rods, triangulation, theodolites all kinds of optical instruments to cite alignments of things. And it was an extremely slow and tedious procedure, and full of errors, to make measurements of the deformation of the earth that way.

But starting in the 90s, the global positioning satellite system was deployed, and then it was possible to make the measurement with just a GPS receiver. Just put a good GPS receiver someplace, and you can see how all the motions are happening in the West Coast. That's become a very valuable tool. When we do have an earthquake, you can see the change in the motions caused by the earthquake, and then furthermore, these days, we've got something we call space-based radar altimetry. It's a ranging system based on radar, and it allows us to see how the distances between a single satellite have been affected by an earthquake and it allows us to take a photograph of the deformation around important earthquakes. That's been a very important new development as well.

ZIERLER: Another collaboration with Hiroo, your work on microscopic and macroscopic physics of earthquakes. What do those terms mean? What are microscopic and macroscopic physics of an earthquake?

HEATON: That's getting into some deeper stuff. The question is about the physics of earthquakes. And I think I've mentioned this before, one of the big issues in earthquakes is, they're not nearly as violent as you might guess if you'd never seen an earthquake, except you went to a laboratory, and you measured the stresses and energy stored in a piece of rock at the depths at which earthquakes happen. An earthquake happens typically 10 to 20 kilometers deep in the earth, and the pressures on the rock at those depths are enormous just from the weight of the rocks above. The pressures would be, like, three times the pressure you'd get from the equivalent in water just because rocks are much denser. If you take two pieces of material, jam them together, and try to make them slide, the friction between those materials is so high that you end up with enormous forces necessary to make them slide.

If you put that kind of enormous force on the rock, it strains, and it stores up energy elastically. The amount of energy that's stored is staggeringly big. In the laboratory, when you do this, if you hold together, and you force it to slide, when it does slide, the forces are so big that it's quite violent, even with little, tiny bits of material that you're forcing. You don't put your fingers near these experiments because they're big, powerful machines. You'll lose your fingers. If that was what the physics of an earthquake was, you'd have to scale that little cube of rock up to something that was as big as Southern California, bigger than Mount Wilson. And when it went, it would be so violent, none of us would survive the earthquake. It wouldn't matter if you were outside, the ground would accelerate so violently–earthquakes are powerful, but they're nothing like that. The question is, what's really happening with earthquakes and the scaling of earthquakes that when we get earth-sized failures of these materials, why do they end up looking the way they do?

And this has been a 50-year puzzle in the earth science community. We call it the stress paradox. Why are the inferred stresses in earthquakes so much lower than what we expect from laboratory measurements? As far as I'm concerned, in the field in general, it's still mostly confusion about how that works. There's this paper about microscopic and macroscopic physics of earthquakes, and that largely deals with that subject. It's a good and important paper, but I don't think it's the end of the story. I think I've figured out how this works, and I've written it into about a 120-page manuscript, but the problem is, I know that if I try to publish the manuscript in the known journals, I'll die before it gets published. It's just too fantastical for most reviewers. These days, you need four positive reviews to get something published. Even if you've got three, it's almost impossible to get four. It's really hard to make breakthroughs these days because there's a lot of inertia to our system. I put it in my class notes because nobody can stop me. [Laugh]

ZIERLER: I'm curious if the Northridge earthquake was something of interest to you even several years after 1994.

HEATON: Northridge really is kind of an interesting case. Prior to Northridge, there was the 1971 San Fernando earthquake, and it's a virtual twin of the Northridge earthquake, but Northridge was a little bit further to the west. And the San Fernando dipped under the San Gabriel Mountains, and the Northridge earthquake dipped the opposite direction. But they're very similar earthquakes, and very similar things happened. But why were they 23 years apart? Why 23 years? It's kind of crazy. We had two magnitude 6.7s, and the mountains are all across the front of that Los Angeles basin. Could we have bigger earthquakes on those structures? The San Fernando earthquake, after it happened, we could see the fault, and we realized we should've seen the San Fernando Fault before it happened. It turns out, people hadn't. But if they'd really looked, it's kind of obvious.

But after the Northridge earthquake, we never saw the fault. It didn't really come to the surface. And if the earthquake hadn't happened, we wouldn't have known there was a fault there. It just kind of shows you that you could have an earthquake of that size and importance on a fault that you don't even know is there. This kind of phenomenon happens in our business fairly often. There was a very famous earthquake in Tokyo in 1923, magnitude 8, directly beneath the city. No surface fault. People even now kind of argue about what actually made that 1923 earthquake. If you look in the estimates of what could happen in Tokyo for a big earthquake, people say, "We could have an earthquake as big as 1923." Duh, of course, it already happened.

How about something bigger? "No, we don't see how you could put that"–but you don't understand anything about 1923, so how you could say it couldn't be bigger, there's no sound science to make statements like that. All you really know is 1923 happened, and you shouldn't be too shocked if something bigger happens. We need to understand that our understanding is not adequate to put limits on that. When we go to design this campus, what should be built on this campus? People build these models, they say, "Here are the faults, here are the ground motions that could happen." My feeling is that virtually anything could happen. You don't know these things. At some point, you have to acknowledge that there are a lot of things we don't understand.

ZIERLER: That's part of science, though.

HEATON: Well, sure. A good scientist knows what they don't know. You realize that's just the way it is, and you proceed from there. If you're an engineer, you've got to proceed based on the understanding that there are things you don't know. But life has to proceed anyway. Right now, people are making decisions, "Well, my crazy design is okay because I can design for the projected ground motions." My feeling is that you could make a not-crazy design that could handle much bigger ground motions, so why not do that? Why make this crazy design based on thinking you know what's going to happen, when in fact, usually, we don't?

ZIERLER: That's a great place to pick up for next time. We'll start in the early 2000s and bring the story right up to the present.

[End of Recording]

ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Tuesday, May 10, 2022. I'm delighted to be back once again with Professor Thomas Heaton. Tom, great to be with you again. Thanks for joining me.

HEATON: Thank you, good morning.

ZIERLER: To start in the early 2000s, we've talked a lot about your collaborations with Hiroo Kanamori, tell me about specifically your collaboration, Microscopic and Macroscopic Physics of Earthquakes, and how generally how your and Hiroo's areas of expertise have worked in such a complementary way over the years.

HEATON: There's quite a longstanding mystery in the earthquake business that is about energy and earthquakes. In particular, it's been known for years that the average change in stress in earthquakes is a very small number compared to the stresses that exist deep in the earth. Depths of earthquakes, which are typically on the order of 12 kilometers or so, the pressure of the weight of the rocks above is enormous. It's the pressures of the very deepest parts of the oceans. In laboratory situations, when you try to make rocks slide with such huge forces pressing them together, the forces are enormous. Typically, about two-thirds of the confining pressure. When the rocks actually break in the laboratory and start to slide, it's quite a violent thing that's seen.

And if you scaled up what you saw in the laboratory up to the scale of the earth and real earthquakes, then the earthquakes would be so violent they would basically kill all of us. If you were anywhere close to the earthquake, the motions of the ground itself would be enough to actually break your legs and kill you. We couldn't survive them. And earthquakes don't look anything like that. The paper with Kanamori is an attempt to describe and solve that problem. It's been around for quite some time that earth scientists have viewed the stresses in earthquakes are much smaller than what is seen in the laboratory. And the paper with Kanamori really describes that problem in tremendous detail and makes a very strong case that earthquakes don't look like what we've seen in the laboratory situation. And it does it by tracking energy in systems.

It's an important paper, and it's a follow-up to a paper that was written in 1985, a very famous paper by Kanamori and Anderson about stress change and earthquakes. The paper that Hiroo and I wrote kind of adds in information about the apparent friction on faults that wasn't included in the Kanamori and Anderson paper. To be completely frank, I think there are still some missing pieces to the puzzle in the Kanamori and Heaton paper. I've written what I think are the missing pieces in a set of notes that are my class notes for my Engineering Seismology class because I spent two years trying to put together the rest of the puzzle as far as I'm concerned, and I realized that if I tried to publish what I wrote, I would never get it through the reviewers. It's so different from current practice and theory.

ZIERLER: What's different about it?

HEATON: It relies on chaotic physics, so it says that the process consists of slip pulses, which are very localized, solitary waves of slip, which was the hypothesis I made in a paper back in 1990, and the paper I did in 1990 was a very controversial paper. And I suggested in that paper that the slip pulses had profound implications for that stress state and friction in the crust. And I still believe that, but it turns out it's an incredibly difficult mechanics and mathematics problem to solve the way the systems work. It requires computers that haven't been invented yet to do the numerical calculations, or else it requires some very clever applied mathematics that hasn't been invented yet that I know of. But I think I can guess the answer to this problem. But in science, guessing the answer is important, but often, it's hard to convince anyone else that your guess is correct. I'm quite convinced that my hypothesis is correct. But getting other people to even understand what it is I'm trying to hypothesize is often difficult. Hiroo and I had many, many conversations about what I was hypothesizing, and I think he often felt that I was outside the bounds of normal physics, geophysics at least. It's not that exotic in the world of theoretical physics, not that I'm a great theoretical physicist.

But I think it's the only solution to the problem. It says that if you have a pulse running through a system, it cannot run steadily. If everything's uniform in the system, and there's a pulse, it turns out as the pulses run, they just get exponentially bigger or exponentially smaller, they cannot run in a steady fashion. There are no steady solutions, which is part of the reason it's such a difficult problem. Usually, we like to find steady solutions and study them, but they don't exist for this problem. Some people have concluded that means that the problem is nonsensical, that it doesn't have any real-world applications, but it turns out that in chaotic physics, those kinds of problems are just like what I'm describing, and the key to the problem is that if you have a system that has this kind of characteristic, if you let it evolve over time, the initial conditions become very complex. They look like what we call a fractal. Once the fractal is something that evolves from the dynamics of the system, if we describe all the stresses in the crust as fractals, I think it explains most of our observations. But it's an extremely difficult thing to see because rocks are not transparent.

To see what's going on deep in the earth, you have to either drill into the earth or look at rocks that have had the tops eroded so you can see what they looked like when they were actually deforming. And we see very complex structures in the crust. Most earth scientists would say it's the structures themselves, the complexity–"I've got granite here, basalt here, I've got sandstone in different places." Very complicated-looking geometries. Most earth scientists would say it's the complexity of the geometry that's most important. But in the slip pulse view of things, it says that the stress is extremely low when the fault is moving quickly, and it's extremely high when it's not, so it has these very rapid transitions from very high stresses to extremely low stresses. And those transitions are moving through the material. It turns out that same thing shows up in mechanics of materials in very small-scale dislocations that work through materials in materials science studies. They're a notoriously difficult problem. But they leave behind extremely complex stress in the system, and it's that extremely complex stress that I think is kind of the missing piece of the puzzle.

It turns out that in things like steel, if you want to know about how steel yields, you have to know what's happened to it in the past. You can't just take a piece of steel and ask, "What's it made of?" and know what its properties are. If somebody's going to make a strain-hardened steel or a special kind of sword, they often heat treat the steel, or they often beat it with a hammer and bend it. All those mechanical things they do to the steel change its properties. It turns out there are dozens of different kinds of steels that have virtually the same molecular composition, virtually identical elastic properties, but the way they fail, some of them bend like a paperclip, some snap, some are brittle, some are ductile, some are very high strength before they fail, some fail at fairly low stresses, and the only difference between all those materials are the micro-stresses inside the materials. This is well-known in metallurgy, but in the earth sciences, this is not well-known at all. Nobody's ever been able to see inside of the earth with enough detail to see what exactly is going on. What I wrote is a conjecture that says that the earth has those kinds of properties in it, and that's why we have the characteristics of earthquakes that we see. It explains the statistics of big and small earthquakes, and it explains a number of observations of stress that we have. But like I say, trying to get published what I just told you, I just know it's much easier for me to kind of self-publish it at Caltech.

ZIERLER: What does that tell us about the culture of seismology these days, some of the orthodoxies that cannot be violated, at least through publication?

HEATON: When I joined [the field of] seismology in 1972, it was much smaller.

ZIERLER: Was it more freewheeling intellectually as well?

HEATON: It was. People were asking really big questions. It had just gone through the plate tectonics revolution. It's interesting that Caltech in the 30s, 40s, and especially the 50s, Gutenberg in particular had been more or less a supporter of the Wegner's continental drift theory, which was kind of poo-pooed by lots of other people. There was another famous geophysicist, Sir Harold Jeffreys at Cambridge, and Jeffreys said, "Look at the seismic waves going through the earth. Look at the mountains standing up. The earth's a solid. How could things drift through the earth?" He thought it was crazy. Gutenberg called it the elastoplastic earth. At that time, plasticity was part of materials science being explored at Caltech.

I'm not sure whether Gutenberg was actually talking to Caltech engineering people about this, but people at Caltech engineering knew about this, and somehow Gutenberg knew about it, too. But at any rate, after Wegner's continental drift, it kind of got rejected, especially because Jeffreys was so strident in his rejection of those ideas. Then, in the 60s, at other institutions, oceanographic institutions, people found evidence that the ocean floors were evolving with time, that there was spreading of the ocean floors. And that kind of nailed the problem and said, "The continents are just rafting around in this ocean floor environment." Suddenly, all kinds of observations fell into place because of the continental drift idea, and plate tectonics was invented, and it explained all kinds of things. All it needed was this great idea, and everything fell together, and it changed everything. It was an enormous revolution. I remember Caltech wasn't really part of the plate tectonics revolution.

At that time, the United States was trying to go to the moon. At Caltech, people were trying to think of the big problem to work on, and they said, "Well, we've got JPL up here. Let's work on geology of the moon, the planets." Caltech had developed a big planetary sciences group, and a lot of the work on various geological sciences, especially geochemistry, was lunar work. Not in the Seismo Lab, but Caltech in general was working on lunar things, especially Gerry Wasserburg, who had something he called the lunatic lab, I forget what he called it. He worked on geochemistry to age parts of the moon. That was another big revolution, finding the age and origin of the moon. That's a pretty fundamental unsolved kind of problem. At the same time, there was a suggestion that we would be able to predict earthquakes in the 70s, especially the Chinese and the Russians, who were both writing papers about precursors to earthquakes and predicting earthquakes. This was an enormous new idea, and people didn't really understand earthquakes very well at all. The notion that we could somehow predict them was very appealing, and a big national program was started to do exactly that. But we found that we didn't understand the system very well at all.

A variety of geophysicists wrote papers about how this whole system worked, but certainly nobody was thinking about chaotic fractal systems or even solitary pulses of slip going through earthquakes. We were just at the beginning of knowing how to characterize and understand earthquakes. Big questions were being asked at that time, and eventually, there were kind of classes–science breaks into schools of thought, and there's a large group of people that were really fixated on this question of the change in stress on the fault due to an earthquake. They viewed that the stress was at some level before the earthquake, and then it reached some sort of a limit yield stress and would just suddenly drop down to a new sliding friction level. And that was called the stress drop, and everything was dictated by that change in stress.

But it turns out that if you view it in the view that there are these pulses of slip going through the material, the change in stress is locally extremely complex. It can go up, it can go down, and when you're done, you just have some sort of very complicated change in stress, but it's got an average value where it goes down because it has to release energy out of the system. The pulses, as they travel through such a complicated system, actually carry energy from one part of the earth to another. If you do an energy balance on the problem, the pulse turns out to be a very fundamental part of the dynamics of the system. But to be honest, nobody in our business talks about pulse energy but probably me and my ex-student, Ahmed Elbanna, who wrote a paper on pulse energy. But nobody else even uses the concept.

ZIERLER: To go back to this idea that there's a gulf between laboratory experiments in earthquakes and earthquakes in the real world, is that suggestive of the distinction in the title of microscopic and macroscopic earthquakes? Does that get to what you were talking about?

HEATON: I think that's right. Microscopic, what we see in the laboratory, versus what we observe macroscopically, earthquakes, there's a big difference between those two things, and it's important to understand that. I think that's well-described in the Kanamori and Heaton paper. But I don't think we quite got all the pieces together to explain why that difference is there.

ZIERLER: Around this time, what's happening with the Southern California Seismic Network? What are some of the developments?

HEATON: The time of that paper, about 2000, was when we were new fleshing out the new broadband network, which was this seismic network that everything was recorded on a computer at the sites, then the network turned from a network of just sensors into a network of computers that had sensors on them. The sensors were now many, many times more capable than what we'd had previously, with which we could only measure a very small range of amplitudes, whether they went on a piece of paper or later through telephone lines. You could only send a very small range of small to big numbers on a telephone line the way we did it. We used something called frequency-modulated telemetry, which was very inefficient. Once everything went to sending actual digital information, then we could send very small numbers and very large numbers through the same telephone lines, and telephone lines became so much more capable that we could now send enormously greater bandwidth of data through our systems. I think I've said before, that really revolutionized our networks.

But when I studied the earthquakes, I was especially interested in earthquakes that were magnitude-6-plus and knowing what the evolution of the slip over time was on the faults. And to do that well, you need instruments in close to the earthquake, and you need sort of a dozen of them within 20 kilometers of the earthquake to see it in different places and directions. Kanamori took a little different tact on the whole problem. He was especially interested in big earthquakes. His research especially focused on subduction earthquakes, which are where the ocean's crust is returned back into the earth. It's big in Japan, very natural, and most of the earth's shaking energy in earthquakes comes from those big subduction earthquakes. In the 70s, he made this extension to the magnitude scale based on energy that was radiated by the earthquakes.

And that was another enormous revolution in our business because prior to that time, the San Francisco earthquake in 1906 was considered to be bigger than the 1964 Alaskan earthquake, but when you actually calculated up the energy between those two earthquakes, the Alaskan earthquake was probably 100 times bigger than San Francisco. And people hadn't really appreciated that factor of 100, which is enormous. And in particular, Hiroo characterized these really giant earthquakes that dominate the entire global budget of earthquake slips. A lot of his emphasis was on these really big earthquakes. He's so versatile, he's worked on all kinds of things. But his work on these really giant earthquakes was transformative.

ZIERLER: Speaking of big earthquakes, tell me about your work simulating and modeling the magnitude 8.3 Tokachi-Oki earthquake from 2003.

HEATON: I was especially interested in what might happen to tall buildings in big earthquakes. About the time that I joined the Caltech faculty, my research interests really broadened out, and I had to teach and learn more about the physics of buildings. I was in the engineering department, but I started telling people I was a building physicist, whatever the heck that means. I don't design buildings, and the engineering community generally didn't recognize that I was one of the club. But they did recognize that I was learning and doing some interesting things about buildings. And I kind of focused my understanding about buildings based on my physics training, although eventually, I learned a lot of engineering things, too, which affected my physics understanding. It's funny how things kind of feed back on themselves. When I joined the Caltech faculty, we were worried about what might happen if we had a Northridge-like earthquake in downtown Los Angeles, one slightly bigger than Northridge.

John Hall and I collaborated especially on that, and we concluded that the buildings that were being designed at that time might collapse in that kind of scenario. I started working on the problem of big earthquakes and buildings when I joined Caltech, and one of the related problems was, where do you find the ground-motion recordings from earthquakes to work on this to make sure that you're grounded in reality? 2003 was a magnitude-8 earthquake in Japan, a subduction earthquake, and it was very well-recorded. By the time of that earthquake, Japan had put out many new instruments in response to the Kobe earthquake in 1995. By the time of the Tokachi-Oki earthquake, they had a very good network in Hokkaidō. A student of mine, Jin Yang, and I worked on processing the Japanese data, and then running those ground-motion records through building designs to see what those ground motions would've done to buildings. It turned out even though they recorded a lot of things, Japan had virtually no tall buildings in Hokkaidō at that time.

ZIERLER: It's a bit of a grim question, but I wonder if you can walk me through the physics and dynamics of tall building collapse in a major earthquake. In other words, with massive shaking that lasts for three to five minutes, I can imagine a few different scenarios, points of failure, of how does a building collapse? What have you found? Where is that point? Is it in the foundation, the upper half?

HEATON: In the engineering world, it's considered natural that in really strong shaking, you would have important damage to a building. You could have a building that goes through an earthquake, and the motions are small enough that everything remains elastic and undamaged, but at some point, the shaking will get big enough that you'll start to do some damage to the building. The argument has been that in building designs, it's okay to have damage in rare earthquakes, it's just not okay to have them actually collapse. What's the difference between damage and collapse? That turns out to be a very tricky subject. In the engineering world, they'd say if a building has small damage, and then you just give it a little bit more shaking, and it collapses, then the system would be called, in the engineering world, brittle, meaning that once it starts to fail, the building kind of shatters.

The most common kind of failure for brittle systems–well, brick buildings are brittle, and the most common failure for a brick building is you've got a vertical wall, and the ground starts to vibrate perpendicular to the wall, and the bricks are not reinforced, so they just kind of fall apart at the base and can't take any tension, that's a brittle building. And that's the most common kind of a failure. Nobody simulates it on a computer. It's really complicated to simulate. They just see it and say, "I know what's happening." The next-most important kind is a kind of building that's got, typically, a concrete column, vertical columns, and concrete floors. The building's called a frame building because it looks like kind of tinker toy frames with concrete floors on it. Often, those frame buildings are made with steel, and the view is that once steel gets up to its yield point, it starts to bend like a paperclip. Then, we call that ductile bending. But in concrete, if you put reinforcing bars in the steel, you can make the concrete ductile, too. You can make it actually bend. And people discovered that in the early part of the 20th century, so they started putting lots of reinforcing bars in the columns longitudinally.

But when the San Fernando earthquake came along, they discovered that even though the steel was in the bars, the concrete would fracture in the columns, then they couldn't take the weight of the building, and the bars would just buckle out. Then, the building would start to collapse one floor on top of the other. In a concrete building with a bunch of concrete floors that are heavy, once you start to lose the columns, it just turns into a pancake stack of floor slabs, and anybody who's in there won't survive. That's called non-ductile concrete buildings. Those were first discovered in the San Fernando earthquake in 1971. The engineering community made some changes to the building code to change the way the reinforcing is put inside of concrete columns and to make the system more ductile.

If you've got a brittle concrete column, once you fracture the column, it just falls vertically, one on top of the other. If you've got a ductile concrete column, it doesn't just fall apart like that, but if you displace the top of the building enough relative to the bottom of the building, the weight of gravity is trying to put a bend on the bottom of the building, and the columns cannot resist that bending force, and the building just folds over sideways, falls over like a tree, basically. That's called a side-sway mechanism. That's a completely different kind of description of how it collapses. One just falls vertically, the other falls over sideways. If you're trying to understand the physics of what's going to happen to buildings, you've got to be able to track that and predict it on a computer. It turns out, that's a pretty difficult problem to actually solve.

ZIERLER: A research project a little closer to home, the results of the Millikan Library Forced Vibration Testing, what was that about?

HEATON: Millikan [now Caltech Hall] is just a fascinating building. Of course, on our Caltech campus, it's an anomaly of a building. It's our only tall building. Most Caltech buildings built from the 30s to the 60s have lots of concrete walls. If you walk through the campus, for all of our older buildings, they're mainly what we call concrete shear wall buildings. Thick walls, not many windows, and very different than the types of high-rise buildings that are usually frame kinds of structures, columns, much lighter weight, more flexible. When the Millikan Library was built, I'm told the donor from the Mudd family told Caltech that they wanted their architect to design the building. We've seen that before. That often is tricky. And the architect wanted this tall building. When you look at that building, it's obvious it was built in the 60s and 70s. But it's got very thick concrete walls. There's no other tall building I know of in California that's built like Millikan. It's extremely stiff and strong.

When it was built, because it was Caltech, it was instrumented, by which I mean part of the engineering thing was, the engineers built what was call a shaker on the roof of the building. It was rotating weights with some motors that would force the roof of the building, and you could measure when the building would go into resonance, how big the resonance was, what the motions were in other parts of the building by deploying accelerometers in the building. And that was done from the early 70s mostly up until almost present. Now, we've got some permanent seismic stations in the building. Every year, the civil engineering classes used to go out and, as part of the class exercises, would measure the motions of the building. It turned out, when they built the building, its natural frequency was something like 1.65 Hertz. It's very high-frequency for a nine-story building. Then, we had the San Fernando earthquake, and during the earthquake, it shook fairly strongly.

And the natural frequency of the building appears to be 0.9 Hertz. 1.65 down to 0.9, that's an enormous change in frequency. It's as if the stiffness of the building dropped by more than a factor of two during the shaking. You might think there was damage to the building to have such strong change in frequency, but no one could ever find any structural damage to the building. And it's easy to get in and look at all the concrete walls in the building. Just go in the stairwells, and you can see them. The building looks fine. After the earthquake was over, it was measured again, and it went back up again to 1.4 Hertz. Didn't go back up to 1.6. It was now at a new frequency, so something permanently changed. But during the shaking, it was much lower than it was even afterwards. It's changing over time in some complicated way. Then, at some point, I said, "We need to put part of our Seismic Network in Millikan so we can record continuously and process the data from it continuously the same way we do in the Seismic Network. We don't need a separate network for Millikan."

What we discovered was kind of surprising to us. We could pick up the natural frequency of the building just from ambient vibrations, people walking around in the building, because now we had very sensitive instruments. But what really shocked us was, when it rained, the frequency of the building would go up by about 3%. The stiffness of the building went up. And then, it would come back down over the next week after it rained. To be honest, I don't think we ever really figured out why it did that. There's another very curious part of Millikan, which was that when they first installed those shakers there and did those experiments, Paul Jennings, an engineering professor who was sort of running those experiments at Caltech, and Clarence Allen, who was in seismology, were having lunch at the Athenaeum, and Clarence was telling Paul about something very unusual that was showing up on the seismometers in the Mount Wilson station, which was that they see this one-hertz signal showing up just for an hour or so, and then it'd go away. And this one-hertz signal would show up from time to time, and they were kind of curious about it. And Paul realized that when he looked at the time, it was the time when they were actually shaking Millikan Library. When they shook Millikan Library, it showed up on the seismometers at Mount Wilson, which was kind of surprising. There was a fair amount of work trying to calculate the mechanics of small vibrations of Millikan would excite seismometers tens of kilometers away. Later, one of my students and I shook Millikan for hours at a time, and we could see those vibrations all the way to the US-Mexican border.

ZIERLER: With all of that shaking, were the librarians upset that the books were going to fall off the shelves?

HEATON: No, it didn't do that, but it did make people a little seasick. In those days, we would only shake at night. My poor student would run the experiment between midnight and 6 am. But when we did run it for classes, we'd kind of warn people in the building we were going to shake it, and it's pretty mild shaking, but apparently it made people feel a little dizzy to have that kind of shaking going on.

ZIERLER: Tell me about Debra Smith. When did you first meet her, and what have been some of your key points of collaboration?

HEATON: Debra was one of my graduate students. She took some of my classes. She came out of physics from Harvey Mudd, and I think she was attracted to the way I did science. She was originally working with Mike Gurnis on plate tectonics things, and that wasn't going all that well, so she asked if I would be her advisor, and I was. And we especially worked on the problem I was interested in at that time, what if stress was really looking like a fractal? How would it manifest itself in other ways? And she wrote a thesis that created a model of a fractal stress and what the implications would be on seismicity, the orientations of small earthquakes. Nobody was doing anything like that at that time. We wrote a long paper that I think is very revolutionary, but most other people who looked at that paper thought, "What are they doing? Why are they doing this? Where did this model come from in the first place?" For me, I knew why I wanted a fractal model of stress, but I couldn't convince anybody else at the time that this was how things worked. Maybe we'll get there. But when she wrote that paper, it was not very well-read. It was accepted, but not many people used it. They just didn't know what to do with it. But over the years, people are finding that lots of things in that paper are showing up in new datasets.

ZIERLER: It was a bit ahead of its time, it sounds like.

HEATON: I think it's way ahead of its time. I think it's actually a very important paper. But it was way ahead of its time.

ZIERLER: The question you posed in your 2007 paper, Will performance based earthquake engineering break the power law? First, what does that mean, performance-based earthquake engineering?

HEATON: About 1995, almost all building codes in the world, including the United States, were based on what people had seen in past earthquakes. After an earthquake would happen, and there'd be some city shaken by an earthquake, the engineers would go out and see what worked and what didn't work. If it didn't work, they'd change the building code and say, "It would've worked better if I had more braces or if I made the size of the walls bigger," or they'd discuss how to change the rules so that it would've worked better in an earthquake. Then, starting in probably the 90s or so, the earth scientists were now getting more sophisticated at saying, "Here are the kinds of ground motions that can happen in earthquakes," and people were making numerical simulations of how buildings would react to those ground motions, so it was becoming much more quantitative and predictive between, "What's the ground motion?" and, "What will the building do?" as opposed to before, where it was just, "We better make the walls bigger or the like." In the 90s, there was the Pacific Earthquake Engineering Research Center headed at Berkeley, and Caltech was part of it.

And the central focus was, instead of designing the buildings by a set of rules evolved from past earthquakes, they'd build the buildings based on expected ground motions and expected behavior of different designs. It requires a computer model of the building and a knowledge of what future earthquake shaking will be like. Of course, we don't know what future earthquakes will be like, but there's been evolving a very elaborate, predictive model of hazard. It's a hazard model. It says, "Here's the probability you will get this kind of ground shaking," and it tries to incorporate lots of information about geology and seismology into the probabilities that you will get some earthquake ground motion. It's called performance-based because it says, "Instead of just a set of rules, I'm meeting a performance standard," which in most cases is, "I want to require that there's no collapse in the 2,500-year ground motion." At the time the Earthquake Center started and came up with that set of requirements, I was just joining the Caltech faculty in civil engineering, and the Center asked me to be the Caltech representative into their Research Advisory Committee, the committee that was making the decision to go for this performance-based earthquake engineering. I argued that the most important thing we had to do was understand some big earthquakes because the big earthquakes, I said, are the ones that really matter.

These things like San Fernando and Northridge are not the real important earthquakes. You've got to understand ones like 1906 or maybe even worse than 1906. Those are the ones that are going to really matter the most. And if you don't understand them, how can you claim you're designing for the future? From my perspective as a seismologist, I know those big earthquakes have to happen. They're the main actors in plate tectonics. From their perspective, "Oh, those are so rare, I don't really need to worry about them." They said, "1906 is so infrequent, it's not our main concern." I said, "You don't know that until you understand 1906." We had kind of a falling-out then, and they basically threw me off that committee. I can be pretty irritating in that kind of situation.

ZIERLER: Well, you speak your mind.

HEATON: Yes, I do.

ZIERLER: What, so far, is the conclusion? Will it break the power law?

HEATON: I don't think so. Power laws come out of many things, but the most common one people know about is something called the Pareto distribution. It comes out of economics, and it asks where the wealth is in the population. In most statistics distributions, there's a mean of the distribution, and then things die off from the mean. And it says most of the action is near the mean. That's a normal distribution. This power law distribution is a Pareto distribution that says most of the action is at the extremes of the distribution. They're sometimes called fat-tailed distributions. It says that the end of the distribution, the Musks or whoever, have got most of the money, and the other end of the distribution has got nothing. That's a power law distribution. Turns out that a lot of things in earthquakes are power law distributions, where the biggest earthquakes do most of the work. Like Hiroo said, the 1960 earthquake had as much energy released as all the other earthquakes of the 20th century.

One earthquake was the same as all the others combined, just like wealth distributions. It turns out, when you start looking at disasters in general, they're usually described by power laws, whether it's fires, wars, epidemics. If you're trying to describe the statistics of a system that's a power law in the first place, it's really tough. If you're an insurer, and you want to know how much you're going to pay out in life insurance next year, if the insurance is to cover things like auto accidents or heart attacks, all of those events are statistically independent from each other. If I get in an accident, it's not going to cause you to get in an accident, so you can do the statistics using standard probability theory. But if the deaths are caused by epidemics, then you've got to know about the contagion, how schools are, vaccines, you've got to know all these things that really complicate the whole problem. If you're going to predict how many people are going to die in COVID, it's almost impossible to know. It's really complicated. But those big things like COVID might just dwarf everything else.

How many people are going to die in a war? Might be all of us. You just cannot predict that system. And that's an example of a chaotic system. If you're trying to design for the future, and you claim everything is the statistics of heart attacks and auto accidents, but it's really the statistics of wars and epidemics, you're going to really miss out. And that's what that piece is all about. I think what I wrote in there is completely correct, and people should really be talking about it. But in the engineering world–I don't want to talk about it. If you think that it's really a power law like that, then the whole structure that's been built that allows people to build our current city has to be rethought. Nobody wants to rethink it, they're too busy building the buildings.

ZIERLER: I wonder how Jim Beck's expertise in Bayesian inference theory has been an asset for your work in early warning research.

HEATON: We're a small department in civil engineering, so any of our students who go to defend their PhD thesis are going to end up with basically all of us on their committee, which means that any of my students will end up with Jim Beck on their committee. And Jim Beck's a very talented and bright guy, and if you're making a claim about the statistical significance of anything you're doing in science, and all of our science takes some stab at the statistical significance of the study, Jim Beck will always say, "This would be much better if you posed this as a Bayesian inference problem." [Laugh] I just knew that no matter what I was doing, my students would have to answer Jim Beck's questions in their thesis defense about Bayesian inference. And Jim and I are good friends and colleagues, so Jim taught me about Bayesian inference theory because I was on all of his students' committees, and I had to learn about it to know what was going on in their theses.

It became obvious to me, as it is to Jim Beck, that the Bayesian approach to statistics is the only rational way to do statistics. But again, most of the rest of the world doesn't do Bayesian statistics because it makes the problem more difficult to do it that way. In particular, Bayesian statistics says, "You can't just take a bunch of numbers and do statistics if you're going to ask the most likely outcome. If you have enough numbers, you can say that with just the numbers. But if you only have a few numbers and want to know the most likely interpretation of what you're looking at, as human beings, you say, 'Here's the mean of those few numbers, but I also know some other things about this problem. I may have some understanding of looking at other problems that are similar in that there are certain things I expect to happen in this system.'" If I only had one number, I'd say, "Here's the one number I've got, but it's different from what I expect. I'm going to say it's most likely going to be what I expect rather than the one number I just measured."

If I get five numbers, if there's a difference, those five numbers are important enough that they start to change my mind away from what I expect. If I have 1,000 numbers, I better really change my mind and forget my prejudices. Bayesian statistics says, as human beings, we've got some prejudices, which are often valid. And if there are enough numbers, then you don't need prejudices anymore, but if there aren't enough numbers, you better use your prejudices as well. For early warning, we're trying to get information out to people as quickly as possible. An earthquake starts, and only one station starts, and it's only got a few numbers out of that station because it's just started. You're going to get a lot more data later on, but at the beginning, you've got almost no data. You see something and say, "I think it's an earthquake. What is it?" Well, we've got some prejudices about, "This earthquake's in a place where we just had an earthquake. There are aftershocks going on. It's probably another aftershock." That's my prejudice. I don't know for sure until I get more data, but I can build that kind of prejudice into my statistics using Bayesian inference theory. And Jim Beck really helped my students to understand how to do that. And like I said, it complicated their lives, but I think it's the only rational way to do that problem.

ZIERLER: Is using Bayesian inference theory relevant for the real world? In other words, your interest in real-time testing and performance of Early Warning, what's the distinction between the theoretical and laboratory work and what's actually happening with earthquakes?

HEATON: Jim Beck has converted me, so I would say yes, we should use the Bayesian inference. But to be honest, it does complicate the problem. My other colleagues I collaborate with to build this system, mainly out of geophysics, have never been through this experience of having to sit through all these PhDs on Bayesian inference, and they don't know what I'm talking about. And their lives are already complicated. They don't want to know. They say, "I don't need that to solve this problem." But we've run into some parts of the system where we do better if we put in the Bayesian inference part. But it does complicate the system, and it's already complicated enough. There's a fight between, "We want to keep it simple," and, "Do we want the optimal answers?"

ZIERLER: In 2011, were you involved at all in studying the Fukushima disaster?

HEATON: Remember, I had studied the Tokachi-Oki earthquake you asked me about, and I'm reasonably well-known in Japan, probably in some ways, better known in Japan than in the US. And in 2010, there was a conference on long-period ground motions at the University of Tokyo, and I talked about what would happen in very large subduction earthquakes to tall buildings, and it was based largely on that Tokachi-Oki study. And I told the Japanese people at that conference that I thought that the area off of Ibaraki-Oki prefecture, the area just due east of Tokyo, the subduction zone of Japan Trench all the way up to the northern part of the island, that the available data was compatible with having magnitude-9 earthquakes on that section of the subduction zone, and that they should consider that as a possibility. And they were incredulous. The stock thought was, "We only get magnitude-8s here. We've got a long history, and there are no magnitude-9s here." To have me say that was very odd to them. Then, as part of that conference, they took us out to a nuclear power plant on the southern part of Honshu. It's called Hamaoka Nuclear Power Plant.

They were showing us how strong this nuclear power plant was. And nuclear power plants are built like bunkers. They're incredibly stiff, almost indestructible things. And this nuclear power plant's built on a bench just about five meters above sea level, which makes it easier to cool. I said to the lead engineer who was giving the tour, "What's going to happen when a tsunami sweeps across this?" He said, "It's impossible." I said, "You can't be serious. You don't actually think that, do you? Obviously, this place could have a tsunami sweep across here." He said, "That could never happen." I told the people at the conference and the engineer I thought that kind of disaster was obviously a possibility, and they just dismissed it. After the real earthquake happened, the official Japanese stance was, "Nobody could've ever foreseen anything like this." To be honest, Kanamori and I both had discussed this numerous times, including discussions with our Japanese colleagues. But I think they felt there was just nothing they could do with that discussion because there was too much momentum in Japan, and to suggest that someone had seriously considered that in a prior way would've just been a tremendous loss of face for the Japanese community to admit that. I've got the emails, but it's well-buried otherwise.

ZIERLER: We've talked about many detection networks. A new one is the Quake Catcher Network. What is that, and how did that get started?

HEATON: About 2010, people realized that inexpensive, solid-state accelerometers, like are used in gaming devices, had sufficient precision to record the motions in earthquakes. They also control your airbags, and those ones are generally designed for very strong accelerations and not very precise. They really weren't useful for earthquakes. But then, they started making a new generation of similar solid-state accelerometers, meaning they're just on a chip, that were designed for the kinds of motions you make when playing a game, moving your hands around. And those are the kinds of motions you get in the earthquake. They were making them by the millions because once you design it and put it on a fab machine, they can stamp them out in huge numbers. We could buy those controllers standalone for on the order of $35 apiece. And you could record them on a standard PC, and PCs were everywhere. Suddenly, there was the possibility that you could have seismometers built for less than $100 and use people's PCs to record the accelerometers, and then use their internet connections to transmit the data back to look at it.

We thought, "That's great." And around the world, there are lots of people interested in this problem, and we'd get volunteers to record these things. A number of people just thought this was a cool idea. I don't know whether any one person in particular was the first person to have this idea, but there were a couple of young students at Stanford who ran across this idea, and they were so excited about it that when they went on to university research positions, they didn't let go of that idea, and in fact, they asked for startup funds from their universities to get them started using these devices. That was the Quake Catcher Network. That was Elizabeth Cochran in the USGS building at Caltech, so she's a Caltech affiliate. And she worked with Jesse Lawrence at Stanford, and they worked on making these sensors and shipping them out to different parts of the world. And they got some additional funds from NSF to do this. They shipped them down to the earthquakes in New Zealand, and they were fairly successful.

At the same time they were working on that, Monica Kohler, in civil engineering at Caltech, and I got interested in the same idea, and Rob Clayton got very interested in the idea independently, too. We had sort of three groups of people that were all thinking, "This could really revolutionize things." For us at Caltech, we were trying to figure out how we could proceed. And that was the time when Gordon Moore had given that enormous gift to Caltech for kind of unrestricted research. And we put in a proposal between Rob Clayton and Monica, and we also had computer science working on this. Mani Chandy was especially interested in how to do crowdsourced sensing, and this was kind of a perfect problem for it. We got some funding from the Moore Foundation to get this thing going, and we designed a client to go on PCs, and we went out and tried to convince Caltech faculty and alumni to take them home and install them. Then, we went out to schools and tried to get school kids interested in taking them home and installing them. It took a while, maybe a year or so, for us to discover that especially schoolchildren are not very reliable. People who took them home to their PCs, they'd get new PCs, or the operating system would change, various things would change in the system, and it was very difficult to keep the system running. It would deploy, but it wouldn't keep operating.

After maybe five or six years, we decided it was much better for us to operate the system ourselves, which meant we had to have somebody to maintain the system. It was still very inexpensive. The devices were inexpensive, easy to site, just go out. And we especially started to work with Los Angeles School District. This was the Community Seismic Network. You asked at Quake Catchers, which was the Stanford-UC-Riverside network that Jesse and Elizabeth put together. And it went well for a while, but they were young faculty, and eventually it just ran out of steam. The Caltech effort somehow just kept chugging along. We evolved and changed it into this different model, where we were working especially with LA Unified School District, then we were working also with JPL, and JPL helped fund a network of about 150 of these sensors on the JPL campus. And between JPL and the LA USD, we had maybe 700 stations out, which is a large number of stations, more than the stations in our expensive broadband network, so that was the good news in the system. The bad news was that these things only got good signals if the motion was large enough that you could feel it. If it was smaller motions than you could feel, which is what a traditional seismic network usually records, then our network wouldn't pick up anything, so you have to have significant local earthquakes to get any signals.

And we had the bad luck that we started the whole thing during a profound earthquake drought. It's like any drought, hard to predict when it'll be over. The Ridgecrest earthquake was in 2019, and just prior to that, Caltech had been funding a lot of this thing, and it was the Moore Foundation, and then funding came from other places, directors' funds and things like that. It was really becoming problematic to keep the funding going for this network. And we decided we'd run out of money and would have to mothball the entire system. Somehow, Monica found a little extra funding to keep it going for a few extra months, and along came the Ridgecrest earthquake. The motions from Ridgecrest were recorded throughout the Los Angeles Basin by this Community Seismic Network system, and the records are just spectacular. No one had ever seen anything like what we saw on those records. What it shows is that parts of Los Angeles south of downtown have very amplified long-period ground motions in that earthquake, up to a factor of ten larger ground motions for tall buildings down sort of near USC.

People didn't really understand that until this network recorded it in Ridgecrest. That's actually having some real impact in the engineering world. Most of the strong shaking in California has traditionally been recorded by the California Geological Survey in a program called the California Strong-Motion Instrumentation Program, which operates a network of far-more-expensive sensors than what we were using in CSM, and their network would only trigger during an earthquake and wouldn't record all the time like the Community Seismic Network. Recently, at Caltech, in order to see this network [and to] go forward to the future, we collaborated with UCLA on using this network, and the reason is because it's especially exciting to the engineering community, and UCLA's got a big civil engineering department. Caltech's civil engineering department is kind of retiring away, so there's not much in the way of students to work on these problems anymore.

Anyway, we've been collaborating with UCLA, and furthermore, the California Geological Survey has taken note of this system, and they recognize that this system might be their future, so they're providing funding to keep it going as well. I think as we look to our future, it's pretty obvious that these kinds of solid-state sensors are going to be ubiquitous. They'll be in all of our automobiles. Inertial navigation of cars requires both good GPS and good accelerometers to microposition where the cars are. Cars just sitting and idling will be really good seismometers, and they'll just be everywhere. All of our phones nowadays are full of accelerometers, so Google is running a project to harvest the data from cell phones. We're headed into a future where we'll be able to describe the motions everywhere in a city, which will help us do emergency management and help us to understand how the motions vary from one place to another and what controls those motions. It will clearly revolutionize our business.

ZIERLER: In 2014, you took a retrospective approach to Northridge on the 20-year anniversary. Both from civil engineering and seismology perspectives, how had the field changed, and what did we get right back in 1994?

HEATON: I think I was kind of angry at the time when I wrote that. [Laugh] I think mostly, my paper about the retrospective of the 1994 earthquake was about engineering issues. I was upset that in 1994, we were well-aware of the tremendous problem with non-ductile concrete buildings. That's what I mentioned earlier, concrete buildings with inadequate reinforcing in the columns. And we saw collapses of that in 1971 in San Fernando that made the code change. Then, we saw those same kinds of collapses in Northridge, and we realized we kind of dodged a bullet in Northridge because the heavy shaking was out in the wooden houses, the suburbs. You take the heavy shaking we recorded at Northridge and put it downtown, we'd have probably had tens of thousands of fatalities. We had recognized, as a community, these non-ductile concrete buildings for ages, since San Fernando, but nothing had been done about it by the time I wrote that retrospective, although amazingly enough, there is now an ordinance in the city of Los Angeles that people will have to deal with the non-ductile concrete buildings.

It was pushed through by Mayor Garcetti, and it takes a lot of political clout to get that through. Lucy Jones helped a lot in getting that through. It remains to be seen whether they can actually follow up on it because to fix non-ductile concrete buildings is expensive. It's not an easy fix. There will be a lot of complaint about it, I'm sure. There's a time lag in the fix, so it hasn't happened yet. Then, I complained about the brittle weld problem that showed up in Northridge. That's when they first recognized there was a problem with brittle welds. Basically, nothing's been fixed with that problem. But again, the same ordinance that was put into place with the city of LA for non-ductile concrete talks about brittle welds, and it's pushed even further into the future. Unfortunately, people who are in dangerous buildings are typically completely unaware that their building could collapse and kill them.

ZIERLER: Tell me about your work with Sarah Minson and the status of the Shake Alert project.

HEATON: Sarah was a student in geophysics. Not my student, she worked with Mark Simons on especially how to understand the slip in earthquakes, a problem I worked a lot on in my career. She was working on how to interpret GPS data in terms of earthquake slip problems, and I was on her PhD thesis. She's a very bright student, and I've talked with her quite a bit. I'm not sure how it happened, but somehow, she also got the Bayesian inference bug from Jim Beck. A lot of her thesis was how to do this inversion problem using Bayesian inference theory, and I ended up on her thesis committee, and we might've hired her to work on Early Warning when she was first out of grad school. But I encouraged her to consider going to the USGS, which she did, to work on this problem, and we continued to collaborate on how to use Bayesian inference theory in seismic early warning or just how to do early warning in general. Early warning's an interesting problem. It's this fusion between traditional geophysics and engineering because it's not just ideas, it actually has to work. You have to make a system that reliably works, which is more like what you do when you're an engineer. People often ask me the difference between engineering and seismology, and for engineers, it actually has to work.

ZIERLER: How did you get involved in the idea that optical systems could be useful in seismology? In other words, what can a still photo or a video do that a seismometer might not be able to do?

HEATON: Looking at instrumentation and things like buildings, for instance, Millikan has got seismometers on every floor to keep track of all the vibrations and deformations of the building, and when we did the Community Seismic Network, we put stations on every floor of some tall buildings so we could infer how the entire building was moving around, then we make some computer simulation of what it would look like with amplified motions. I thought, "I bet if we had a really good video camera, in a real earthquake, we could just have the video of the building and digitize up all the straight lines in the building, the windows, the corners, and directly see how the building is deforming in an earthquake. I thought that was a cool idea. I went to talk to Mike Brown, an astronomer, about that. Mike had other interests at the time, so it kind of disappeared. There was an ex-Caltech graduate student, Shervin Taghavi, who was interested in how he would use his background in optics. He got his PhD in electrical engineering at Caltech and worked for JPL for a number of years.

He was working on optical systems, and he wanted to use his knowledge to make optical-based seismometers. We have varieties of ways of using optical systems. I won't get into the details here, but I told him I thought we had really good traditional accelerometers to do the job he was thinking about, but it would be hard to use what he was talking about to compete with our existing systems. But I said, "There is this thing that if we could use the videos to see how buildings are deforming in earthquakes, that could really be transformative. It could really change things around." He got excited about that. He thought, "I could do that." He went out, and started to work on it, and came back, and I suggested to him that to test it, maybe he should go down and take some videos of the Vincent Thomas Bridge, which is a big suspension bridge out to Terminal Island that moves around all the time.

He went out with a telephoto lens and took some videos of it. And you can see the bridge moving in the video frame, and every time the bridge would dip down, a truck would go across. And we quickly recognized that the trucks were causing deformation of the bridge. You could see it drop down. And we quickly realized just by looking at those videos that you could infer the relative weight of the trucks going across the bridge. If you drive along the freeway here in Pasadena, the truck lanes are destroyed within a month of them fixing them. Obviously, the trucks are tearing up our freeways. It didn't take much research to find that it's mainly the very heavy trucks that are doing that. We thought, "We could probably spot heavy trucks just by keeping videos on bridges and infer to the people like CHP or Caltrans, "These are the distributions of what the truck weights are, and here are the trucks that are especially heavy." Right now, they measure the trucks with the weigh station, but they only get a very small percentage of the trucks in the weigh stations. It just takes too long, and there are too many trucks.

Furthermore, the truck drivers know where the weigh stations are, and if they're overweight, they typically avoid them. Or they know when they're closed. But if you've got a camera out looking at random bridges, they'd never know. It'd be a much stealthier system. We recognized this could actually have some value to society and could be the basis of a startup company, so now there's a startup company called WeighCam, and it's gotten some funding from the NSF Small Business Grant program. And we're hoping to have our first prototypes available in the next couple of months. Joel Burdick in engineering has been collaborating on this as well. This is a big surprise to me that at the end of my career, I'd have something to do with a startup company, but if we could get overweight trucks off the road, that'd be fine by me.

ZIERLER: Bringing our conversation closer to the present, between your contributions to projects like PhaseLink and working with colleagues like Mani Chandy, what role do you see deep learning and artificial intelligence having for the future of earthquake prediction and early warning?

HEATON: First off, we're awash in data. Now, we've got all of these seismic sensors all over the place sending continuous data. There's no way a human being has got the time and brainpower to deal with that enormous amount of data. There are all kinds of secrets hidden in these continuous datasets, so how do you mine and extract the information you want out of this enormous ocean of data? Data mining. You need techniques that mimic what you would do if you had infinite resources. That's what most of these AI techniques are. You teach them how to do things that you would do, then you say, "Repeat what I would do a kajillion times," using algorithms to capture what you've done. And you don't even have to understand what you've done, you just have to know what your interpretation would be, then you can have the computer pick up and do the same thing you would do. And it's really fascinating to think that you could process so much data. The hard part to me is that it'd be ever so much more satisfying if I understood why I did that in the first place. As a trained seismologist, I know what I'm looking for sort of intuitively.

People have told me, "Pick that. Pick this. Pick that." They taught me to do the interpretation of things, but they never really taught me exactly, "Here's how you know it." It's just an example of something, and your brain recognizes, "I've seen this before." You teach the computer, "I've seen that before," but you don't actually write down and have the computer say, "Take this derivative and add it to this second derivative," or whatever. In essence, it does, but you never get back from the computer the true insight where somebody would write down, "Here's exactly what the computer's doing." It just says, "It does it." When you do a Google search, they'll tell you, "Here's how the algorithm works," but if you said, "Exactly what is it doing?" nobody really knows because the computer's just doing this algorithm. And the algorithm's often very different from the physics you were thinking about in the first place. But the idea that we can just handle these enormous datasets, it will be transformative. Ultimately, we are an observation-driven science.

ZIERLER: Looking at your last batch of graduate students, perhaps from Lucy Yin to Becky Roh, what are the kinds of things that the next up-and-coming generation of seismologists are looking at that might provide a window into where the field is headed?

HEATON: I'd like to think we'll start to see some convergence between the mechanics sciences and geophysics and that we could actually test things. In my worldview, science is about coming up with some creative ideas, and then developing them into what I like to call settled science. Settled science, to me, means it's something where I can write down a set of conditions and equations, and I can solve that system, and I can reliably know that if I give that same set of things to some other sufficiently trained person, they'll get the same answer. It's amazing to me how often that doesn't happen. In a sense, it has to happen in engineering because whatever you're working on has to work. Sometimes the scientists view the engineering as kind of boring because the science has been done, or sometimes the engineers view the science as kind of flaky because nobody's ever shown that it actually works. You've got to get the idea, then shepherd it into it reliably working. That's what our business is all about. It's quite a difficult process.

When you look at the earthquake process, we've got a huge amount of work to do to get it into settled science. I think I've said it before, but it especially irritates me to see people talk about strength of materials in the earth when they don't understand strength of steel. If you can't understand strength of steel, why do you think you could understand anything about strength in the earth? I'm not saying the earth is made out of steel, but it might be that you just don't understand anything about strength in the first place. It amazes me how in the earthquake business, the only tool we've got to look inside the earth is to look for variations in the seismic velocity in the earth. And we have some areas that are faster and slower, and that's kind of like medical imaging. It's relatively known, although there are still some really fundamental problems even there. Then, they come up with fast and slow, which is usually colors, blue and red or whatever, on a map.

And they say, "That's strong. That's weak. That's heavy. That's light." If you tried those same techniques on an experiment in the engineering world, it'd be ludicrous. It just doesn't work. The whole endeavor is not repeatable in that way. I think it's really important for the earth sciences and the engineering sciences to come to some sort of fusion between those two. But for me, it's been a 50-year career. And I've had a 50-year career at a place like Caltech, which is really strong in both of those areas, so that's what's happened in my career, that I see that there are lots of disconnects. And I looked at some of my colleagues at other places, and the engineers would never talk to the seismologists at those schools. And they don't even have an engineering school at many of the other science schools. But if they do have an engineering school, I can guarantee their engineers are not talking to their scientists. It just doesn't happen.

ZIERLER: On that note, we've worked our way right up to the present. For the last part of this excellent series of discussions, I'd like to ask a few retrospective questions about your career, then we'll end looking to the future. First, a touchstone of all of your work, how you've had this simultaneous research agenda in seismology and civil engineering, do you think that's a model for the future? Do you see scholars taking this dual approach? And what might be the best way to get people out of their academic siloes, given the fact that there is so much relationship between these two areas?

HEATON: I think it's really important for people to come out of their silos, but as I just said, it's incredibly challenging because we're all so busy fulfilling our jobs in our silos. To make the connection to people outside–they're busy, too. The language is different. The way it worked for me is, there were some very special people and mentors at Caltech. Caltech's this really small place. I took classes in engineering from people like Jim Knowles, brilliant engineering minds, and they mentored me, and shared with me things they knew, and gave me an opportunity to pick up things I never could've had in some other, bigger place. Caltech is so small that people get to know each other, and the personal connections are just critical. People challenging each other about fundamental ideas, "Why are you doing that? What does it mean that you do that?" Other schools are just too big to do that. It's been a very special time and place for me to pull that off. But to be honest, I think it's going to be even more difficult at Caltech because science continues to grow in breadth, and the mechanics parts of Caltech are becoming less at the forefront with time. And we're kind of evolving towards the medical biological sciences, especially at Caltech. It's been a unique experience for me, but I couldn't necessarily see that other people would be able to reproduce it. And if I had to do it all over again, I think I'd do it, but there are lots of frustrations and difficulties from trying to be in two communities at the same time.

ZIERLER: On that point, a theme we've come back to repeatedly, the orthodoxies in seismology. Among them, where are you most optimistic for change, that people will think more expansively, focus on the problems that you think are most important?

HEATON: We're having this revolution in instrumentation, getting signals at all length scales, and we're getting computers that can deal with these signals at all length scales. And lots of things that in our business have been just called noise. "I want to look at this part of the signal, then there's lots of other stuff in there that I don't understand, and because I don't understand it, I'm just going to throw it out and call it noise." I think now, we'll start to see what it is that we're calling noise, and it's a lot of stuff that's a part of our signal, and we'll find there's a lot of information in that noise. Some of that will be revolutionary, and it will really start to let us see how things are changing at all length scales. As I mentioned before, I think we'll find fractals all over the place, and then we'll have to deal with the question of fractals. What's the physics that makes the fractals in the first place? Hopefully, we'll get some very clever people looking at fundamental problems. You asked me about fundamental problems in the 70s, and it feels like people have been shy about getting back to fundamental problems in the last 20 years. I wish we'd ask more really deep fundamental questions about how things got that way.

ZIERLER: If there is the big one, in Japan, Southern California, or anywhere else, where are you most concerned from a civil engineering problem that the alarm bells you've raised have not been heeded? And to flip that question around, where are you most confident that there have been some advances in the building engineering where we are pretty well-prepared for a massive earthquake in an urban center?

HEATON: It turns out that the old way of doing earthquake engineering, which was the standard codes, we'll go out to see if something didn't work and come up with a new set of rules that make it better, is a remarkably robust way from a societal point of view. Our current set of houses and short buildings is probably very robust, even in very large earthquakes. It's kind of ironic that probably the best set of buildings anywhere are 7-Eleven stores. Nobody spends a lot of money on those, but they're very robust. And it wasn't that hard to come up with systems that did that. But I'm sure we'll see collapse of probably dozens or maybe even hundreds of non-ductile concrete buildings. This collapsed building we just saw in Miami was clearly a non-ductile concrete building. You see just what a horrendous tragedy those can be. I don't even like to think about it. If we saw side-sway collapse of some 80-story buildings, basically, you're talking about a World Trade Center kind of collapse. The World Trade Center was a pretty traumatic experience. To be honest, I don't really like to think about these things. They kind of give you nightmares. I'm not sure what to say. I hope I'm dead by the time that stuff happens. I don't want to ever be in the position to say, "I told you so."

ZIERLER: My historical appreciation of the Seismo Lab going all the way back to the beginning, of all the ways the Seismo Lab has revolutionized the field, what do you see as most important looking to the future? How can the Seismo Lab maintain its leadership and remain an intellectual center for the field?

HEATON: I would say the key to the Seismo Lab is the same as the key to Caltech, which is, find the very most creative minds you can find, and keep them communicating with each other, and encourage people to collaborate across disciplines. I think that's really important. When you look at it from that point of view, you just know that important questions will be asked and important new things will happen. It's almost impossible to tell you what they're going to be. It's going to be somebody very clever asking some really difficult questions, and somebody else in another field saying, "That's kind of a familiar problem to me. Here's how we approached this problem." I think that's how these breakthroughs really happen. This last couple of years has been really tough because our communications have been so disrupted by COVID. It's really a tough time recently. But typically, Caltech works very well, and I hope Caltech just continues to put creativity at the forefront. All this nonsense about social justice–we're not a social justice institution, we're a hard-nosed technical place for real creativity, and that should be our focus. That's what I'd say. Don't lose track of that.

ZIERLER: Finally, for you, last question looking to the future. Now that you're at a point in your career where you can choose to focus on the things that are most interesting and fun to you, what do you want to do that you haven't accomplished yet?

HEATON: [Laugh] When you're at Caltech, so much of the important stuff is not done by us, it's done by faculty together with students, young minds. It's the mentorship between the faculty and some young student who doesn't really know what's going on yet, but the faculty helps to guide the student into doing something really revolutionary. Once you're emeritus faculty, you don't really have students anymore, so we kind of slow down at that point. And that's certainly true for me. I think frankly, for me personally, I hope all the people I've managed to collaborate with over the course of my career will continue to value talking to me because I certainly value them. Those people have been critical for me. And it's a great joy for me to get back with my ex-students or ex-collaborators and just enjoy each other. That's what I'm really looking forward to.

ZIERLER: Tom, it's been a great pleasure spending this time with you. I'm so glad we were able to do this, capture all of your insights and recollections. I'd like to thank you so much.

HEATON: Oh, sure. Thanks.