Anant Agarwal

INTERVIEWER: It's Friday, August 28th, 2015. I'm Chris Boebel

and I'm here at MIT with Anant Agarwal. Anant is a pioneer in computer architecture and a professor of electrical engineering at MIT. He's also the founding CEO of edX. The online learning destination established in 2012 is a joint partnership by MIT and Harvard. Anant taught the very first edX course from MIT on circuits and electronics which drew 155,000 students from 162 countries.

He has also served as the director of CSAIL, the Computer Science and Artificial Intelligence Laboratory at MIT. Anant is also an entrepreneur. He has co-founded and led the development of a number of successful companies, including Tilera, Virtual Machine Works, and Oxygen.

Anant has received multiple recognitions for his teaching and research. He is a member of the National Academy of Engineering, a fellow of the American Academy of Arts and Sciences, and a fellow of the Association of Computer Machinery. He holds a Ph.D from Stanford University and a bachelor's degree from the Indian Institute of Technology at Madras.

So welcome, Anant. Thanks for coming today. Thanks for sitting down with us.

AGARWAL: Oh, thank you for having me.

INTERVIEWER: So in reading your biography, your life story, it's fascinating that there are these two threads that sort of come together, this sort of thread of education and an interest in education. And then, of course, electrical engineering and computer science. And I'm wondering about the origin story for that. Going way back to your earliest memories growing up in India, were they present from the beginning or is it something that you found your way to?

AGARWAL: Well, I found my way into electrical engineering and computer science quite late in life. I grew up in a small town in India on the Arabian sea coast, of the Western coast of India called Mangalore. It was a small port city. Yet, it's Mangalore. I like to joke that it's Bangalore with an M. Bangalore is a much bigger town.

I went to school there, to a public school. To a Jesuit school. And a there, I had the usual public school kind of education in India. And certainly, through high school, I really had not been exposed to electrical engineering or computer sciences. There were certainly no computers back then. This was in the early and mid '70s. There, funnily enough, my ambition was to become a fighter pilot in the Indian Air Force. And it turned out that at the end of your 10th grade, you have to take an exam called the NDA. It was a National Defense Academy exam. And I was all set to go do the exam and try to become a fighter pilot. And there I never heard of IIT. Mangalore was a small town and really, when I was in 10th grade, to my knowledge there had been nobody from Mangalore who'd gone to any of the IITs.

There I was inspired by my math teacher, JL Saldanha. So he was very inspirational and he exposed me to IIT, and he said, hey, look you should try to get into an IIT. I had not heard of IIT before, and then when I learned about it, it was-- it seemed completely unattainable. IIT accepted one in 100 students that applied. And just in comparison, MIT admits about one in 10 students that apply. So this was just extraordinarily out of my reach. And so I thought this was beyond the pale in terms of what was possible.

But he really inspired me to aim high. He would tutor me and help me. And he wouldn't take a penny for it. So he worked with me, asked me to come to his house, and taught me math. Or really help me bone up in math.

The standards of the school-- the school was good, but the public education standards and so on were pretty bad. And the IIT entrance exam was a completely different-- at a different level altogether. So I'm did quite a bit of studying, by the end of 10th standard, I had gotten-- I was getting more and more excited about the IIT.

What happened that year, was the NDA, the Defense Academy exam was changed from being after 10th grade to after 12th grade. And so I was pretty disappointed that I couldn't take the exam to be a fighter pilot. But fortunately for me, or unfortunately for me, I don't know, in the last two years, 11th and 12th standard, my interests shifted to electronics. And then I applied to IIT and I managed to get into IIT Madras. And so really that's how my interests changed, pretty much towards the end of my high school.

INTERVIEWER: Tell me a bit about your family, your parents. I mean, growing up in a small town like that, were they supportive of this dream that seemed unattainable, supportive of education in general?

AGARWAL: My parents were totally supportive. My father was a professor of pathology at the local medical college. It's called the Kasturba Medical College, KMC. My mother-- she had a masters in English, but she was a homemaker and she also started a business in our own home that enabled her to augment my father's income.

So we were a middle class family. There were very heavily into education and really inspired us to really get well-educated. I still remember the days when my father would come home, he would dictate 100 words, and I would have to spell them every single day. So they were certainly very involved in our learning and doing better.

So they were completely supportive. But it pretty much left us to do what we wanted to do. They didn't push us. They really had not heard of IIT themselves, and so it was really up to me to aspire to what I wanted to do and become.

And what was interesting, is even after I got into an IIT, which is Indian Institute of Technology, I remember one day my father was-- as you know, parents like to brag about their kids. My father was bragging to a local-- another friend that oh, our son is going to be going to IIT. And they were quite surprised. They said, whoa, you mean he could not get into the local engineering college. Is that why he's going far away?

And so, in Mangalore, you aspired to go to the local engineering college. And so, they thought I was going there, because I did not quite make the cut to go to the local college. So it was pretty interesting how the whole local feel, certainly in the early to mid '70s was just a completely different culture from where things are likely to be today.

INTERVIEWER: And it's interesting that you say your parents, obviously, were very focused on education, very focused on doing well, but not necessarily pushing you in any direction, not necessarily laying out for you the path that you ended up taking.

AGARWAL: Yeah, our parents did not lay down the path at all, whatsoever. They didn't have a strong feeling one way or the other. They didn't push us. They pretty much gave us space.

And it could be that the times were different. Where we had our space. We enjoyed a lot. Really, our school did not give us any homework. I did not do any homework at home. I would come back and spend my time playing and having fun.

And once I got into a building things, I would do a lot of building fun gizmos and gadgets and so on. But really, there was no organized activity as such outside of school. And our parents did not push us or did not suggest any paths that we go down. They were simply supportive of whatever we want to do. It doesn't help that I was doing quite well in school, and so they didn't feel the need to ask me to work harder or-- they just pretty much let us be.

INTERVIEWER: You mentioned that it was really in high school that you became interested in electronics and building gizmos and gadgets. Are there any strong memories that you have of what sparked that interest or how that came about? Because that seems like a major shift in your life you know. Talk a bit about how that happened?

AGARWAL: You went through middle school and high school and I would be puttering around and building various things. And now that I look back, I like to build things. But it never occurred to me there was a career there. I had wanted to be a fighter pilot all the way growing up, for whatever reason.

And so, I do remember that in sixth grade, an uncle visited me and he gave me a kit. And the kit contained a power supply, contained some copper wire. It contained some magnets. And you could play around with it. And so I remember I tried to build a motor for an electrical motor for one year, and I could never quite do it.

And I finally figured out how to get the things right. There was no web to look up things. And I would go to one of the teachers in my middle school and ask them, how do you do? And of course, nobody had any idea. We learned science in school, of course, but-- so I would putter around and be building various things.

I also remember creating an explosion once. I took a whole bunch of matchstick heads and I made a hole in the mango tree in our backyard, and I filled it up with the powder from the heads of matchsticks. I took a big nail, and I anchored the nail in the hole. And then I took a hammer and whacked it in and there was an explosion. And so I would do stupid things like that, not quite knowing what I was doing.

So I did have fun building things like this, but really I wasn't thinking of it as something that was a precursor to a career or any such thing.

INTERVIEWER: So you applied to the Indian Institute of Technology, IIT, and you've got in. And obviously it you know, it has a phenomenal global reputation. Talk about that experience, making that transition, and what it was like getting an undergraduate education there.

AGAWARL: So the transition to IIT was actually very scary for me. Because when I went to school, I did very well in school and everything was very easy. In fact, frankly, way too easy. I really did not find a need to study or do any homework. The standards were just not great. The teachers were good but the standards were just not-- I would say-- up to a world-class standards in math or physics and in other subjects.

I somehow managed to sneak into an IIT, IIT Madras. And there, I still remember, physics was taught-- physics was calculus-based. And they expected you to know calculus as it applied to physics, and I had no clue what was going on. And so in the first mid-term, I remember there were about 200 students in the freshman-- 200 or 300 students in the freshman class taking physics. And in the first mid-term, two students failed. And I was one of them.

And so that was very scary time for me. I go there from a small town background not having learned calculus-based physics and so on. So today, when students come into MIT, they've all taken advanced placement calculus and BC Calc in high school and they have learned advance physics based on AP physics. I went in there-- had not applied. I knew a little bit of calculus, but I hadn't applied it to physics. And so I was one of two that failed. And it was a very harrowing experience when I wondered, what the hell am I doing. What the heck am I doing in this place?

But I worked up-- worked real hard to catch up. And eventually, things got better as I managed to catch up and learn that, oh, the way you solve the problem is by applying calculus. And here I was, trying to discover calculus in the midterm exam by trying to subtract the velocities at different times, and tried to use the deltas in the velocities to compute what they were asking for. I had no idea how to do it. When a simple derivative would have given me what I needed.

So fortunately, I caught up. Being with all the other students was phenomenal. Just the whole group of students, they've all become close friends for life. It was frankly, the best five years of my life may have been my five years at IIT. Where all the students were not just bright, but also interested in all kinds of things.

One of my friends was the national chess champion. And another friend who was a national swimmer. There was another student who could solve the Times crossword puzzles in five minutes or less. Everybody was amazing at a variety of things, public speakers, drama, games, sports, really good at a number of things. So it wasn't just about work and study.

And we learned a lot from each other and people helped each other. It was just a remarkable, a very remarkable experience as we grew up and learned with a group a phenomenal students. At IIT, if you looked at the teaching and so on, research-- I just tended not to be, and still don't tend to be a very heavy research oriented place that focuses more on undergraduate education and some on graduate education.

And as you might imagine, the quality of the teaching was-- there were great teachers and there were some not-so-good teachers. But really the group of students that went through it together were a absolutely amazing in how we learned from each other and provided support. And they've all become dear friends for life.

And what is interesting is that even today, in Silicon Valley, there have been book after book written that talk about how this diaspora from IITs were one significant aspect of the whole startup culture. There are studies that showed that a significant fraction of startups in California had at least one IIT founder in them.

INTERVIEWER: So during your time at IIT, what were the subjects they drew your interest? How did those interests develop over your undergraduate career? And talk about how that led you to the work that you started out doing once you came to the US.

AGARWAL: Sure. In India, interestingly enough, you're admitted into college, whether it's a regional college or an IIT based on merit. Pure and simple. It's just based on your-- at regional college it's based on your grades, politics, physics, chemistry, and math into engineering.

And IIT, it's an entrance exam. And they just rank you from one. In my year, I think a couple hundred thousand students took the exam and the top 2000 made it into an IIT. And they rank you from 1 to 2000. And you can just pick your subject based on your rank.

And so for me, about that time, I had wanted to do electronics. There was really no computer science at that time. Electronics was the most popular field and also the most sought after. And so they went by your grade in the entrance exam and I didn't quite make the cut for electronics at IIT Madras. But I made a second cut which would give me mechanical engineering.

But I really wanted to do electrical engineering, and it turned out that at the end of the first year and if you were among the top what, two or three students in your entire class, they would let you-- it was sort of like a wild card. You had this wild card opportunity and they let you switch to any branch that you liked, and so I was lucky enough to win the wild card slot, and as part of that, I switched from mechanical to electronics. And I started in electronics following that.

And even there, in electronics-- the teachers in electronics, the Professor Rao and Professor Reddy were exceptional. Two of the best teachers of circuits and electronics I have ever seen in my life. The pair had even invented a new op app. It was called the Rao and Reddy op app. Great teachers. So I learned a lot from them and I got more excited about electronics.

I did a little bit of computer science on the side in my freshman year at IIT. There was this computer center with this big IBM mainframe. And I did learn to program with punch cards. We used punch cards and I did some programs. I remember my first program was computing the value e using a series calculation.

We were doing small programs with punch cards. We really did not get into computers, but certainly I was doing my IIT study in electronics.

INTERVIEWER: So that was your first introduction to computers at IIT? You had not worked with computers before that? Or what was your awareness of computing and electrical engineering up to that point?

AGARWAL: My awareness of computing was certainly zero. To me, my first notion of a computer was in my first year at IIT and where I used punch cards to program e on the mainframe. Where you turned in your deck and you got the results back the next day. That was my first, absolutely first introduction to a computer. I had not heard of computers before. I may have known what a computer was, but really, it wasn't something that was even on my radar screen, so to speak.

Electrical engineering-- I had built small little gizmos like motors and had built small, started to create the various gadgets. Invent funky stuff and so on. Using electrical stuff and so I knew some electrical stuff.

But I really had no experience with the electronics. I had never used or never heard of a transistor, as such. I just knew about electrical stuff, so anything I did was with magnets and wire. I would build things like speakers and so on and so forth. But really no, absolutely no, connection to electronics at that time.

INTERVIEWER: So I suspect that soon we'll be talking about teaching styles, methodologies of effectiveness of teaching. Tell me a little bit about the teaching style that was prevalent at IIT at the time you were there. You mentioned having a couple of amazing professors. What was the predominant style of teaching and what was your thinking at the time about whether or not it was effective?

AGARWAL: So teaching style-- whether it was in my relatively modest high school in Mangalore or the teaching style at IIT or for that matter the teaching style at MIT-- was identical. There's really no difference in teaching style. So as an example, in IIT, we had a smaller group of students. So in electronics, the entire class of electronics was 30 students. So our classes had 30 students in them and that's it for the entire year. For that year, there were 30 students.

And so we had desks in the class and we sat down. Each of us had a little desk with a place to put our books. The teacher would lecture on a blackboard and we did not get homeworks as such, because there was no concept of the teaching assistants. The professors would lecture and we would take notes. And that's it.

Each semester-- it was a semester system-- we had three exams and one final exam. And so, the three quizzes, if you will, were offered at three times in the year. And before each quiz, they would give you a what they call a tutorial sheet. And the tutorial sheet had a set of problems. There were no solutions given. They were not graded. And you would just practice. And then you would go and take the exam.

So that was a teaching style very much like MIT with the exception that at MIT we give our students weekly homeworks. And we have a lot more resources to hire teaching assistants where the teaching assistants and graders grade the homeworks, which is very expensive and give it back to students. And IITs-- we did not have that.

So we did not have at least, when I was there in the late '70s, there was no homework. There was no concept of homework. So we would just prepare for exams and go in and take the exams and get graded. And there were labs, weekly labs in certain subjects. But otherwise, the teaching style was virtually identical. The professor would stand in front of the class and they would lecture to the students and write on a blackboard. That was the predominant major teaching part.

There's no one on one tutorials. There was not-- didn't have the small group tutorials. No grading of homeworks. Nothing. Just the teaching was a professor lectured to the students.

That was the majority of teaching, which frankly, is the way I learned in high school since I went to first year of my high school, first year of my schooling in 1964 or 1965 and that's the system I had there. We all sat in front of the teacher in neat little rows, and t the teacher would lecture at us and wrote things on the board. Same thing at IIT and the same thing at MIT for most of my 27-year career today in the mid 2015 at MIT.

INTERVIEWER: So a while ago, you mentioned the diaspora from my IIT. And I was wondering if you could talk about your own decision to come to the United States to attend Stanford for graduate school. How did that come about? What was your thinking at the time? I'm not sure if you were familiar with the US or if you had traveled in the US up to that point, but tell us the story.

AGARWAL: So in those days, IIT was a five-year program. I remember there was a debate about whether IITs were going to go to a four-year program. And a lot of the people felt that if would be a disaster if it went from five years to four years. So it was a fire-year program.

So by the end-- the middle to the end of the fourth year-- we have to make a decision about what do we do after IIT. And from IIT, there's a significant fraction of the students would go abroad for higher studies, for a master's and a PhD. And so I decided that I wanted to pursue a career in research. I had not decided that I wanted to teach, but I wanted to pursue a career in research, for which you needed to get a PhD. So I applied to a number of schools around the world and I was lucky enough to get into Stanford.

It is interesting that the one school that has turned me down at every step has been Berkeley. I love Berkeley, but I applied to Berkeley. They turned me down then. I applied to Berkeley for a faculty position. They turned me down again. And so I've had a string of turn-downs from Berkeley in my career, interestingly enough, but I got into Stanford. And there, I started in electrical engineering at Stanford, not in computation.

INTERVIEWER: So had you had much experience with the US? Was this a big life changing decision, obviously, to come here? And once you were here, tell me a bit about the experience of studying at Stanford in California-- in what was not yet Silicon Valley, but that was the moment when it was all emerging, right?

AGARWAL: Right, so I flew to Stanford from India in 1982. And a lot of firsts for me. I had never been out of the country before, so it was the first time out of the country. And in fact, it was my second time I was flying, even. In my fourth year, I had managed to publish a paper at the International Conference on the acoustic speech and signal processing on a speech recognition system for Hindi digits. And I did at that point, aboard my first plane ride flying from India to Paris.

But this was my second time flying on a plane and went to Stanford. And first time living out of the country. Fortunately, I knew-- I mean, I spoke English and at IIT, we used US textbooks and so on. And certainly the level of the IIT education, the rigor and so on, was certainly compatible to the rigor of engineering education in the US. It's just that we did not have homeworks, which was quite interesting. So it's not clear how much we learned, but it was rigorous, nonetheless.

And so when I came to Stanford, I began taking graduate level courses. The transition was pretty straightforward. I worked with Professor Thomas Kailath, who was a research assistant for him in the field of information and decision systems. I switched fields over time, but that's where I started.

INTERVIEWER: It must have been exciting to come down in that area at that time. I mean, that was again the computer PC revolution had begun, but it was still really kind of just beginning in the early 1980s and Stanford was right in the middle of it What was the environment like? And what was the interest that drew you and how did that interest evolve while you were at Stanford?

AGARWAL: Those were really exciting times. I took a class on VLSI, so in fact, chip design and chips. But really, what built Silicon Valley? It's not Software Valley. It's Silicon Valley. And so the chips and silicon was what really propelled it.

And to some of the pioneers in Silicon at Stanford at that time, some of the first design tools we're built at Stanford at that time. MIT had also been building a few things. Chris Terman at MIT had built a simulator. And there were other people at MIT as well, but certainly Stanford was one of the leaders in the field and I was able to work with some of the people building the first CAD tools for chip design, the people doing pioneering work in devices and so on at Stanford.

So it was really exciting being exposed to real leaders in the field. And at Stanford, as I took a VLSI class, I got really excited. I think for the first time in my life, I got really, really excited about a field. And so then I asked around, hey, I'd like to work, do my thesis and build a career in chip design.

And it turned out that two of the professors, Newkirk and Matthews, who I liked a lot and whose class I took, they were going to be leaving Stanford, unfortunately. And so I asked around to say, OK who else could I work with on to the systems level part of chip design, not the device physics part, but the system level design of chips.

And turned out that I was seeking a professor. One professor told me, there's this new professor in this building. He couldn't quite recall his name, and then he said, oh yeah, it's John Hennessy. And he's just joined Stanford. He's doing something with computers and so on and something with a chip. Why don't you go talk to him ?

I walked into John Hennessy's office. And he'd been at Stanford-- he didn't have tenure at that point-- for maybe four years. And I asked him up if I could be-- join his group and being on RA for his group. So he interviewed me and then he gave me an RA and he asked me to build-- So his group had been building the MIPS chip. The MIPS chip became a revolutionary concept.

And he asked me to build the MIPS computer. And so this was my first hardware project, and as part of the group, I built the MIPS computer, which actually were the first MIPS computer. This computer still hangs outside my wall.

And those were incredible days. And we didn't realize that they were doing and using systems that were the first of their kind in the world. So, for example, in the hardware lab at Stanford, I used a multibus card cage. And it was a card cage into which I prototyped, I built a multibus card in which I prototyped the MIPS computer. And what is really interesting was, someone later told me that out of the same card cage, that one of the early founders of Sun had prototyped one of the very early Sun systems before he went out to become one of the early founders of Sun.

And so it was curious we happened to be using the same system. And what I ended up building was actually the first MIPS computer. RISC and MIPS became household names after that, but certainly in those days it was certainly not a household name and a lot of the things we were building were things that would become very well known later.

INTERVIEWER: What was it that first excited you about chip design? I mean, of all of the research issues you could have taken on, of all the directions your research career could have gone, was there something that really spoke to you about that particular subject and that particular challenge?

AGARWAL: I think what excited me was before that, I was an analog guy. I would do differential equations. I would build up circuits for 405 transistors, and do math on paper and do math with matrices and very analog mathematics. And I could do small things. The hard-to-make things work. I built a filter, it took forever to make the filter work and noise-- it was really painstaking.

When I was exposed to digital design and VLSI systems, it completely blew my mind that you could build these little chips with-- in those days, thousands of transistors, even tens of thousands of transistors-- and you could build very complicated systems. And it's very easy. It's unbelievable how easy it was to do it, whereby following and adhering to some principles, you could build very complicated systems with using tools and digital technology where the chips would have a power and performance that was completely, absolutely unimaginable in the analog space. It completely blew my mind, because I came from the analog space.

And so that's when I said, I want to be building these things. I want to be building these chips and I certainly felt that they were going to change the world. And so, that's why I got into it where things that mere humans could put together could have such incredible potency and impact.

INTERVIEWER: So we've been talking about the origins of your research career. What about an interest in teaching? How did that develop? When did it develop? You said, you came to graduate school really thinking you were training to be a researcher, but clearly your interest expanded at some point.

AGARWAL: My life itself-- I sort of took things as they came at me, and things up just changed over time. At Stanford, I was a research assistant all the five years I was there. I didn't teach ever. I never taught that class ever. I did enjoy public speaking and I did enjoy explaining things to people. And even at IIT when I would explain something to somebody else, I thought I was pretty good at explaining things and mentoring people and helping them understand things.

And so, I certainly liked that aspect, but I hadn't thought of class as such. So when I finished my PhD, at that point, I had to make a decision as to whether to go into a teaching career or a research career. And by which point I felt that I wanted to go into a professorial carrier.

Because as a professor, I felt I could do a number of things. I could start startup companies. I could do research. I could teach. I would have resources to do things. And so I felt that if I went into a teaching career, I could be able to do all of the above.

And that's when I applied for a professorship at a number of universities and as I mentioned earlier, Berkeley turned me down. They turned me down as a graduate student as well. So then I applied to a number of schools and I was fortunate enough to get an offer from MIT.

And I really hadn't known much of what MIT was doing in building computers. Certainly from where I was at Stanford, a lot of the work and the energy seem to be coming out of Berkeley and Stanford. And so when I came here, I was up very pleasantly surprised and excited by the cool work that was happening here the work that Steve Ward was doing and the way he thought about things. And Arvind was doing some of his work.

And so what I discovered was at MIT, the thinking was very different and very far out. And I thought this was the place to be. But all of these times I was still very research-focused. I mean, teaching was-- I still wasn't thinking about jobs and so on, as teaching jobs. So really, I was coming here for research and working with students and working with colleagues and others on building cool things and inventing new stuff.

Teaching was something you did, certainly for me at that time. I still remember, when I had to teach my first class, and it's still this way in all our universities. I was told, OK, you're teaching this class. You go teach it. And we don't give people any training, anything, and of course, you do what everyone's been doing forever. And so, I went to my class and I stood there, and for the first time, I felt like a deer in front of headlights.

And it was pretty scary. Here I am in front of MIT students, MIT undergraduate students, you know, creme de la creme, bright kids, and so on. And here I am. It was pretty scary as you teach in front of the students. But you build up confidence over time. But certainly I felt like a deer in front of headlights. I had never taught before. I had never formerly been trained in teaching.

And certainly I would be petrified if somebody had recorded my first years of teaching to my later years of teaching. And I'm sure I learned on the job, but it's not clear. And I certainly saw my ratings go up over time, so I knew I must have been learning the job.

But even there, teaching was never considered the main thing that you did. It was mainly the research. And education and teaching itself as a primary focus of what I did certainly came to me a lot later.

INTERVIEWER: Yeah, I would like to get back to your research in a moment, but I'm curious about just following up on the teaching. I mean, you've won every teaching award under the sun at this point. So how does that develop? Going from someone just saying, go teach. no training and being stuck in that moment, to really developing quite, presumably, a love for it and also certainly, a lot of talent for it. Were there key moments in that journey?

AGARWAL: So there, I think it wasn't studied or it wasn't something that I thought too deeply about. But I found I enjoyed teaching. I didn't enjoy certain parts of teaching like preparing for the lecture or integrating. There were certainly some aspects that were painful, but I really enjoyed the act of teaching where you stood in front of a class and you give a lecture.

And the part that really enjoyed the most was where they would ask you questions and so on. And you answered questions. I find I really, really enjoyed that.

And I think the key was the realization that-- I began to say that teaching is theater. Teaching is theater and to really view it as-- to really give a lot of importance to it, to really be passionate about it, to care deeply about your audience and your students, and you'll do fine. And not to overthink it, but be more organized and really view it like theater.

So as an example, in the very first year that I taught, I would simply go in cold and I would write on the whiteboard. But in my later years, when I was lecturing a big class, I would actually orchestrate how many white boards of material I would write. I wouldn't even plan in my own mind which part of the whiteboard I would write different things.

And so I would plan things, so we used to have sliding blackboards in the 10-250 class or one of the other classes. And I would keep even the layout of the blackboards in mind and I would design my writing on the whiteboards, so that it would be easier for students to comprehend.

The goal was to take out any overhead that was unnecessary. Any unnecessary complexity introduced by the geometry of the boards, and so I would draw pictures on one side and make sure that it would be easy to refer to another board and I would stop and I would make sure that the board was well laid out.

So I would really practice in my own mind as I went through, not the material as such, but how things would lay out. So really, this is almost like choreographing a play, where the way-- you wouldn't even think in my own mind, where would I stand when I said certain things? How would I--

I would have demonstrations. Frankly, I spent the most amount of effort on the demos in 6.002. I think we had some spectacular demos. There's one demo-- it's on YouTube. It's very popular on YouTube involving a chain saw dance that my TA and I did. And the MIT video team they recorded it, and we would dance to Men In Black with chainsaws.

And the main thing was to make teaching fun. And my philosophy is that if you're passionate about it, if you view it as theater, and if you could excite and engage the students, then you simply get out of the way. They'll learn and be fine. But you have to excite the students and get them to want to learn the material. Once you get there, then you can just leave it alone.

INTERVIEWER: That's a wonderful encapsulation of a really powerful teaching philosophy. You spend you spent a lot of time, obviously now thinking about effective teaching, effective learning, how does learning happen, what are the barriers to learning-- at that point, where you thinking in a more organized way about pedagogy beyond the teaching as theater, or was that an interest that came later.

AGARWAL: So I was definitely thinking about teaching pedagogy and certainly in the context of the circuits course, where I found the circuit and electronics material very exciting. And I was in computer science at that time. And one day, in the late '90s, Paul Penfield, who was a department head, came to me and he said, hey Anant. I want you to rethink how we teach circuits.

And so, frankly, when he walked into my office, I said to myself, uh-oh. When the department head walks into a professor's office, looking very grim, you say, hmm, what did I do now? When he comes in and he says, Anant, I need to ask you something. I said, uh-oh. And then he said, I want to take over the circuits course and rethink it.

And the first question that went to my mind is that, are you in the right place. This is computer science. I teach architecture. I taught programming classes and systems classes. But the only circuits that I had done was as an undergraduate, 20 years earlier. I took a couple of circuits classes and that's it. So why me?

And so I'm not sure what Paul had in his mind, but he said, look, Anant, I really want you to do this. And for some reason, I've always been a little bit crazy in my life and when something fun and crazy shows up, scarily enough, I rarely turn something like that down. And so, I tried. I said, hmm, interesting.

And he indicated how the enrollments were dropping, the ratings were not very good, the students didn't like the material. And as I said, let me take a look. It was really surprising that, why were students not liking the material? Why was there not engagement?

So I jumped into it. And my colleague, Jeff Lang and I, we worked on it together. So we looked at the material, and then it became very clear to me that we were still teaching circuits in a very '60s style.

The world had completely moved to MOSFET transistors, but we were teaching circuits using bipolar transistors which were rarely used. There were all kinds of fantastic, innovative applications of circuits that we were using in everyday life, and the course wasn't at any point, never making the connection to those fantastic things that students could do from those. So the content was great and correct and rigorous, but it was missing a lot of the scaffolding, and connections, and excitement, and applications, and in some cases, pedagogy as well. The way certain things were taught, I felt were made unnecessarily complicated.

My style of teaching is I like to excite the students first. Show them an application first, excite them, try to mesmerize them, show them a problem, get them to want to solve the problem, and show them a challenge. And then say, OK, how the heck do you do this? And then once they're excited, once they're sitting at the edge of the seats wanting to know, gee, this looks like magic, how did it happen? Then you go down to how it works.

And at that point, once you've got their attention, it doesn't matter whether you're throwing differential equations at them or deep theory at them, you've got them hooked. But if you start with the complicated theory, and you build up to the app, the students are asleep. They're gone. But the time you come to the exciting app, you have two people left in your class, so we did several things.

So one is we brought in a digital while we taught analog circuits. We showed that the digital world was a key application of analog. We drew the connection to some key application areas.

Second, we built some spectacular demonstrations for every class. Every single class in circuits now had a spectacular demonstration. So we would start every class with a demo, usually saying, huh, how do you think this happens? How do you think this works? And then we go into the foundations and theory.

The concepts were the same, but we really taught differently. So we got them excited. We got them hooked. So those are two or three of the big things. And then we made the course a lot more contemporary. We began talking about many more contemporary technologies and showing the connection to the modern world. So I would say, that these were the things that made the course exciting and certainly the students felt that. I mean, they certainly felt interested. And before we knew it, my colleague and I had wrote a textbook in circuits and electronics.

And so that journey-- we started that in, oh, around '98. And it's 2015, and I still teach the electronic circuits class online. But it started almost as a challenge and almost what seemed like mistaken identity. Certainly when it started for me.

INTERVIEWER: So I, of course, want to spend some time speaking about the development of that online class, but I also don't want to forget that at the same time, you were pursuing a very rich research career at MIT. I was wondering if you could talk a little bit about your directions in research, the kinds of problems that you're really excited about engaging with, and the spin outs, the companies that ended up resulting from some of that work.

AGARWAL: MIT has been a great place. It's been a fantastic playground where you could do all kinds of stuff. And so for me, it was just an absolute perfect choice of career, where I could combine my love for teaching, for research, for starting companies, for entrepreneurialism. It just gave me a perfect cauldron where I could explore all of these things. I don't think there's any other place like MIT, maybe a couple other universities where you can get the same kind of feel-- maybe Stanford, maybe Berkeley-- where you get this perfect cauldron of people that think big, people are crazy, and do crazy stuff, and teach, and start companies. It was just an absolutely fantastic environment that I had.

In the research area, for my work with John Hennessy at Stanford, when I was building computers and computer chips, I continued that work at MIT. And built many new kinds of computers at MIT. We built a computer-- the first chip we built was a chip called Sparkle. Sparkle was the world's first multi-threaded microprocessor. Previously, microprocessors ran one thread, but this was the first chip that could run multiple threads into weave. It was the world's first multi-threaded processor.

And we introduced for the first time the concept of block multi-threading or switch on event multi-threading. There was no name for that concept. We named it. And today, virtually every microprocessor has that concept of threading and switching threads efficiently built into it.

We built a machine called Alewife, which was a parallel computer. And we built technologies for maintaining the coherence of data. Will invented new schemes to build up ways to scale to many large numbers of threads, large numbers, of course. And this was the first scalable cache coherence machine that we built, where the coherent scheme was arbitrarily scalable that had not been done before.

We built a scheme called limitless directories. Previously, this concept called directories that let you scale to a given extent. But limitless directories let you scale virtually forever. So we built the first computer that could scale arbitrarily in terms of maintaining the coherence of data in a parallel computer. So that was Alewife.