The HLF Portraits: Hennessy, John L.

Video in TIB AV-Portal: The HLF Portraits: Hennessy, John L.

Formal Metadata

The HLF Portraits: Hennessy, John L.
Title of Series
No Open Access License:
German copyright law applies. This film may be used for your own use but it may not be distributed via the internet or passed on to external parties.
Release Date

Content Metadata

Subject Area
The Heidelberg Laureate Forum Foundation presents the HLF Portraits: John L. Hennessy; ACM A.M. Turing Award, 2017 Recipients of the the Abel Prize, the ACM A.M. Turing Award, the ACM Prize in Computing, the Fields Medal and the Nevanlinna Prize in discussion with Marc Pachter, Director Emeritus National Portrait Gallery, Smithsonian Institute, about their lives, their research, their careers and the circumstances that led to the awards. Video interviews produced for the Heidelberg Laureate Forum Foundation by the Berlin photographer Peter Badge. The opinions expressed in this video do not necessarily reflect the views of the Heidelberg Laureate Forum Foundation or any other person or associated institution involved in the making and distribution of the video. Background: The Heidelberg Laureate Forum Foundation (HLFF) annually organizes the Heidelberg Laureate Forum (HLF), which is a networking event for mathematicians and computer scientists from all over the world. The HLFF was established and is funded by the German foundation the Klaus Tschira Stiftung (KTS), which promotes natural sciences, mathematics and computer science. The HLF is strongly supported by the award-granting institutions, the Association for Computing Machinery (ACM: ACM A.M. Turing Award, ACM Prize in Computing), the International Mathematical Union (IMU: Fields Medal, Nevanlinna Prize), and the Norwegian Academy of Science and Letters (DNVA: Abel Prize). The Scientific Partners of the HLFF are the Heidelberg Institute for Theoretical Studies (HITS) and Heidelberg University.
Domain name Group action Inheritance (object-oriented programming) Length Real number Bit Calculus Shape (magazine) Line (geometry) Student's t-test 2 (number) Number Goodness of fit Mathematics Internet forum Video game Right angle Musical ensemble Quicksort Algebra Reading (process) Social class Physical system Geometry
Focus (optics) INTEGRAL State of matter Multiplication sign Projective plane Virtual machine 1 (number) Set (mathematics) Sound effect Bit Computer Computer programming Field (computer science) Neuroinformatik Data mining Moore's law Different (Kate Ryan album) Term (mathematics) Video game Energy level Quicksort Game theory Social class
Point (geometry) Trail Service (economics) Direction (geometry) Multiplication sign Set (mathematics) Student's t-test Computer programming Field (computer science) Neuroinformatik Number Normal operator Mathematics Computer configuration Computer hardware Programmable read-only memory Energy level Differential equation Endliche Modelltheorie Computer engineering Hill differential equation Mainframe computer Area Projective plane Calculus Degree (graph theory) Software Universe (mathematics) Computer science Quicksort
NP-hard Axiom of choice Group action Presentation of a group State of matter Multiplication sign Decision theory Real-time operating system Parameter (computer programming) Computer programming Formal language Neuroinformatik Different (Kate Ryan album) Bus (computing) Physical system Area High-level programming language Bit Special unitary group Flow separation Degree (graph theory) Microprocessor Type theory Arithmetic mean Process (computing) Spacetime Point (geometry) Game controller Observational study Translation (relic) Student's t-test Distance Field (computer science) Machine vision Hypothesis Number Goodness of fit Latent heat Population density Average Energy level Task (computing) MIDI Assembly language Key (cryptography) Plastikkarte Limit (category theory) Word Personal digital assistant Thermal radiation Network topology Universe (mathematics) Video game Family
State observer Group action Digital electronics Code Multiplication sign Decision theory Direction (geometry) 1 (number) Set (mathematics) Disk read-and-write head Mereology Information technology consulting Computer programming Neuroinformatik Mathematics Forest ARPANET Damping Physical system Programming language Clark, Jim Moment (mathematics) Bit Lattice (order) Surface of revolution Computer Flow separation Microprocessor Arithmetic mean Process (computing) Order (biology) Computer science Right angle Quicksort Arithmetic progression Geometry Point (geometry) Transformation (genetics) Characteristic polynomial Virtual machine Translation (relic) Control flow Student's t-test Coprocessor Field (computer science) Machine vision Element (mathematics) Revision control Prototype Latent heat Computer hardware Reduction of order Operating system Boundary value problem Energy level Nichtlineares Gleichungssystem Immersion (album) Interface (computing) Projective plane Limit (category theory) Compiler Faculty (division) Personal computer Software Universe (mathematics) Video game Near-ring
Point (geometry) Trajectory Multiplication sign Decision theory Execution unit Online help Student's t-test Event horizon Coprocessor Field (computer science) Neuroinformatik Number Product (business) Different (Kate Ryan album) Computer hardware Authorization Software framework Hydraulic jump Form (programming) Physical system Block (periodic table) Moment (mathematics) Projective plane Interactive television Computer simulation Planning Bit Basis <Mathematik> Faculty (division) Microprocessor Degree (graph theory) Proof theory Process (computing) Personal digital assistant Network topology Universe (mathematics) Computer science Right angle Family Resultant
Complex (psychology) Group action Building Java applet State of matter Multiplication sign Decision theory System administrator Computer programming Neuroinformatik Subset Mathematics Endliche Modelltheorie Logic gate Area Shared memory Special unitary group Computer Surface of revolution Flow separation Microprocessor Dean number Degree (graph theory) Process (computing) Tower Computer science Right angle Arithmetic progression Point (geometry) Transformation (genetics) Virtual machine Parallel computing Student's t-test Field (computer science) McCarthy, John Internetworking Term (mathematics) Energy level Data structure Hydraulic jump Shift operator Scaling (geometry) Key (cryptography) Artificial neural network Forcing (mathematics) Uniqueness quantification Interactive television Denial-of-service attack Faculty (division) Software Network topology Universe (mathematics) Speech synthesis Family
Point (geometry) Divisor Real number Multiplication sign Device driver Student's t-test Reinforcement learning Field (computer science) Power (physics) Neuroinformatik Number Optical disc drive Frequency Different (Kate Ryan album) Internetworking Boundary value problem Software testing Physical system God Task (computing) Area Algorithm Artificial neural network Mathematical analysis High-level programming language Sound effect Cartesian coordinate system Surface of revolution Flow separation Microprocessor Digital photography Word Process (computing) Internet forum Arithmetic progression
professor place me at your origins even before you tell me something about your parents my parents came from a long group of Brooklynites they grew up in Brooklyn they both grew up there they got to know each other sort of high school years their parents knew knew each other they started dating my father served in the Second World War came back went to college after that my mother went to college and began a teaching career as a middle school teacher they got married lived initially in New York City later on moved out to Long Island my father was an engineer and worked in the aerospace industry as it exploded after the launch of Sputnik good good start were you born in I was born in New York City but moved out to Long Island shortly thereafter in mind when I was about four three four where are you placed in the sibling group first first at six first of six yep okay so all the others come along in the Long Island they come along with my long I'm parrot yeah that's your background in Long Island yep okay I'm gonna ask a question I often ask of scientist pushy parents or not you know what do they have in mind for you I'd say what I would say is intellectually engaged parents both my parents were very big readers my father is an engineer you know had a real background in mathematics and they were engaged parents they always wanted you to do your best and and and push you to to do more but not necessarily against your own will you don't remember they're saying this is what we hope it's not playing ball my father would he would he would often ask to check our homework particularly a math of science but he had this phenomena that he would give me back he would say there's a there's one problem that's wrong on this but he wouldn't tell me which one that forced you to kind of go back and review your work which of course young people are often careless doing right right large number of books around the house lots of books around the house my first reading engagement was to read through the entire junior encyclopedia and and then later on I expanded my my reading my father was a big reader in history uh-huh and my mother in fiction so I got exposed to both it's kind of interesting we will talk at length about that yet but actually when you were presiding over Stanford you took as almost as great an interest in the arts as in the sciences so that sounds like it has something to do with the kind of culture you grew up in it did and certainly life experiences where a lot of those things outside of the technical domains become important leadership role okay well along the way we might go there again I assume they put you the system put you in public school or public school to start I I was a bit of a mare do well I didn't work very hard through to my high school years I it was easy it came easily good I could skip the homework and still aced the exam right so I was not very engaged I was sort of bored lovely and then finally my parents decided they need a ninya change and they sent me to a new private parochial school that was just opening huh so I was in the first class of a school that eventually would have twenty four hundred kids in but there were only 600 in the first class right and I had an encounter which is really a life-changing encounter with a math teacher who happened to be a nurse line nun from Ireland and I still remember she it was it was parent-teacher conference and in high school parent-teacher conference the student is there as well as the parent and the teacher and she said I remember I remember to this day she said John has a very fine mind but he has a lazy mind and she woke me up and I realized I wasn't working very hard then she challenged me she said if you want I'll let you take both algebra and geometry in parallel and you'll get a year ahead in math and that way you'll be able to take more advanced calculus and she you know it was your challenge step up and do it she got you and she got me and I took both courses in parallel I got 98 on one and 96 on the other so it was a thing good shape and I like to talk about names recorded of
these people do you remember and yes sister Melba that was your name sister and it made a big difference in my life it really got me really engaged for the first time and in studying and intellectual pursuits beyond the sort of usual things right I certainly had a chemistry set and did all those sorts of things but this got me engaged in schoolwork okay what about another
mentor there or are you now self propelled in terms of effects of your direction that that got me going later in high school at they those days you didn't have access to any kind of computer in the school so instead they had these time sharing machines that you used remotely so there was a small computer Club that had just warmed remember we were the first the first class in this school so we were the ones who were forming all these and that got me interested in in computing early on and that became my focus I went on to do a science project with a colleague of mine to build a machine to play tic-tac-toe which we built out of spare relays that we bought you know surplus kind of thing and built this whole thing up and of course tic-tac-toe is not a very complicated game but what most people don't realize is it's pretty easy to win if you may if the other player makes one stupid move you lose so our tic-tac-toe machine would never lose and it would often beat people because they'd make the wrong move on their first move at the game so how old are you at this play at that by then I'm 17 probably 16 17 okay as you proceed it then maybe even now in retrospect what's the state of the field then in computer what oh very early on very low levels of integration you could do some programming but I'm the machines that we would use remotely because they were not local you had to dial up and you know had less computing power than your your watch probably today and so it's very early very primitive very basic you program in basic or even Fortran 2 which is before Fortran 4 emerged and there aren't really any formal courses it's sort of learn on your own a bit because there's no there aren't yet there isn't yet a curriculum around it so if you you may have in fact done if you went to
your father or the school counselor at this point and said my next stage in
university I would like to do computers what would they have said to you or what did they say so I still remember my father said do hardware not this software stuff as an engineer having come up through the electronics industry for him hardware was king of the hill but I detected that the swing was beginning to software becoming much more important and it was it was it was beginning right were you able to actually enroll in a computer directed program when you went to university by the way where did you go I I ended up going to Villanova you could you could not major in computer science as an undergraduate in the in the 1970s when I entered you could you could do a track and electrical engineering that had a computer option so we had a computer engineering option which is what I ended up pursuing you still had programming courses we we had a programming course and we had a mainframe a 360 model 30 at the University and I got very intrigued by that so we would we cut a deal we would do consulting services helping other students with their programming and learning to program and mentoring them in problem solving sessions oh and then we would go on weekends and on evenings we would get the computer to ourselves after their normal operation was up and so it was an opportunity to go a lot further than the course allowed you to go early on was the cohort that you were in important to you I mean were there more than you invested in this future yeah so I had a in both both at the high school level and college level I had one good buddy who was really interested in computing in the same way I was and we would hang out together and do various things and and work on various computing things together so far you're probably not surprising the world you're fascinated with the computers but your father's an electrical engineer and you're opting for electrical engineering with a computer option correctly so you're gonna be an electrical engineer I started out as an electrical engineer and then I as I got more and more interested in in computer science which was really really developing and standing alone as a field getting a phrase even in the area it was it was in the air and in fact Stanford's computer science department had been formed but they were graduate only programs okay so the phrase was in the air and I got more and more interested in it and decided two things one that I wanted to pursue a PhD I had had a mentor in college that had got me involved in you know undergraduate research project and so I got very intrigued with that and thought I'd like to pursue a PhD and I decided I wanted to go do it in computer science okay so that was the direction I headed how do you do that how do you make that happen so the first thing I had to go I then looked at the requirements and I noticed that their most computer science departments wanted a set of courses in mathematics that were different than what a EE would normally have okay so they were finite mathematics rather than continuous time the kinds of things calculus and differential equations that a EE would have so I placed myself in a number of math courses and I infected I went into one math course I was the only non math major in the course so it really it really challenged me because this is material I hadn't had it was very different than the sort of mathematics that they teach on Electrical Engineering but I enjoyed it and I and I learned the material and that launched me off in the direction of going for pH computer science ok I'm gonna let you go for that PhD okay where you gonna go so I finished my undergraduate degree in three and a half years partly driven by the fact that my longtime girlfriend since our senior prom in high school we had been separated we we had met each other and formed the relationship but we had already decided to go to schools that were about three or four hundred miles apart right with and she was in a very remote place near the Canadian
border the only way to get there was by it took me literally about 20 hours of bus travel to get there so we wanted to kind of get our lodge back together I finished my undergraduate degree so I started looking to see if I could find an institution that would MIT me in the middle of the year yes in my graduate program rather than have to wait till the next September and I had found two institutions that would do that and one of them offered me financial aid and that was Stony Brook and so my wife was my then soon-to-be wife it was doing student teaching on Long Island so I took the offer from Stony Brook back to Long Island back to Long Island itself where she was doing her student teaching okay give me some sense it's a funny word to use maybe of the competence of Stony Brook at this point in the field you're going to become so it's a it's a rising department they had made a number of strong recruits in fact the person would go on to be my PhD advisor Dickie Burt's was originally an electrical engineer himself who actually came to Stanford in a special program that the National Science Foundation had to take people who were interested in computing and give them basically a postdoctoral experience yes that would enable them to move from one field into the computing field oh it's enlightening and he had come to Stanford and done this and then gone back to to Stony Brook where he was then had been made chair of the department when I arrived actually so it's a rising department at the time New York was investing in their heavily in their research institutions and the department was recruiting really stellar students late 70s or that now it's 1974 okay start mid seventies and I had truly a very fortunate experience I was there about three months when a scientist from Brookhaven National Lab who also had an a teaching appointment at the University an adjunct appointment walked in with what became my PhD process so we can't make this up make me so he walks in and what's the the field is very young yeah he's got a problem he's trying to monitor people who are working in areas that that where there could be radiation danger of radiation exposure and he's trying to monitor them in one way you you one thing that happens to people that have long-term low-level radiation exposure as they lose bone density so one way you can scan them with a low-dose x-ray but of course you don't want to give them any more x-ray than you have to right so you have to control this x-ray and do the scan under type control well they didn't have devices to do this microprocessors were just arriving they were just arriving for bit 8-bit microprocessors extremely primitive slow but they could be used for these kinds of laboratory control kinds of things which is what really need to do in this case he walked in and said you know we need a programming system a language and a programming system for doing these kinds of real-time things and there's nothing meaning that people had to program in assembly language and it was difficult of messy and error-prone so he said why don't you think about building a programming system and a language that would work for these kinds of real-time things so that was a brilliant concept remember computer science was still a very young field and so the distance from graduate studies to the edge of the frontier was not very far you could get there very quickly lots of unexplored space so I started working on that that problem and it turned out that just as I was finish my thesis in the last year so let's say 76 I still have another year to go before I'm finished right two of the major leaders in the field start working in the same area and published papers in the area and the interest level just explodes in this area so that that was a stroke of luck I mean it was a good thesis topic but the fact that several other major people who are considered leaders in the field started working on it well that's in the field but what are you doing that isn't duplicating or adding out there that's correct so I was ahead of them by a year too that certainly helped I was more focused on some of the on what we call what what's called in the business hard real-time meaning meaning you have to do a certain task in a certain amount of time and they were focused largely on soft real-time which means on average you have to do so many things in so much time but you don't have to there's not a hard spec a hard limitation for each what it so mine was a little it was a it was enough different to make it interesting and the fact that the fact that these two other people went and began working on it to leaders in the field just reinforced the importance of the area and the importance of thinking about what microprocessors were going to do to the computing field is there a way of describing and these are complex issues but the key insight that you brought to the discussion yeah I think the key insight was let's use a high-level language to describe what the key parameters are so that when we say this task has to be completed in this
time and then let's use sophisticated translation techniques for implementing this so that we can guarantee that these specifications will be met and that was the key idea so who's noticing you or how are you now sorting out the next days so I my first paper you know in a real-time computing conference it's focused on these kinds of issues go to New Orleans with my adviser and actually my wife then were married as graduate student I was married to her and she she comes with me that's my first conference presentation they select the paper for publication in Journal as one of the best conference papers so it gets published in the journal and then I get ready to start applying for jobs fair enough yeah um where do you stumble into so I apply I wasn't sure what kind of reception I would get I decided I want to be an academic okay I didn't want to go to the searched laboratory or industry okay I decided I want to be an academic so I apply to lots of different places and I start interviewing and I end up getting a bunch of offers pretty quickly first place I interview was the University of Iowa they actually had a group working in similar area but I had a lot of other interesting possibilities the very last place I interviewed was Stanford really partly because we're on the quarter system so our interview process a little late but by then I think it was my 14th or 15th visit which was a remarkable process because when you go to get to visit these Department it's not only do they interview you but you get a vision of what's going on in the field and what people are doing that's really an education itself it is an education so but it also meant that by the time I gave that talk at Stanford I've been asked to every question Under the Sun so I could really give the talk really well I had rehearsed it well it had been given so many times I could answer questions well and I remember we when we came out to California we flew back to New York and I brought my wife with me because I said well you know you have to at least see California if we're gonna move across the country both of us from big families that were very East Coast centric you better see it we flew back to New York in March and it landed and it was sleeting and I remember we both looked at each other and said if I get an offer from California we're moving to where the palm trees are and sure enough I got the offer from Stanford and I never looked back I'm good obviously Stanford is such very much the theater of your life before we see that theater I want to roll back to what I should have asked you before and that is let's pursue this question of decisions for industry rather than academic what is really implicated or at that point what you thought was the consequence of going one way or the other yeah I for me it was two things first of all I'd like the intellectual vitality around a university and what happens with students constantly changing and things but the other big piece for me was I loved teaching I had fallen in love with it as an undergraduate when I got to do some teaching of students in the introductory programming courses and then I got to do some it as a as a teaching assistant as a graduate student and I I loved it I loved explaining things to people and seeing them catch on these ideas and so that for me I think was probably the trump card it made me decide I want it's very interesting because in these many interviews I've now done the temperaments of course of people are so critical to the choices they make and some spent their whole life finding a situation where they didn't have students yeah so they could just talk to themselves and nobody else you're clearly not I like both I love my research career and I love my teaching and in some sense you're combining the two I combined the two things I love to do the most okay now I'm gonna ask the same thing I asked before about slowly bringing that and that is your it's after what's the state of the game at
this point at Stanford who's there what did the direction yes so I came out computer science was a very small department I still remember the interview Don Knuth was on my interview as one of the fathers of our field deed I've endured it and he I still remember the question I asked him I said Don how are you so productive and he said to me I don't watch TV and it was I I was not a big TV watcher either at that point I was a pretty intense worker as a graduate student but it was a good it was a good momento and you know he he and I share a lot of things I think a lot of people who are very productive do they love what they do it's not work it's joy to get up and do what you do we'll leave you're lucky enough yeah to be doing it you're getting to do what you really love it's it's a blessing so so that that was a so Stanford is in a interesting place the field of computer science is growing very quickly Stanford has a graduate only program and the computer science department had grown out of our math department so it was in humanities and Sciences small very small 14 faculty probably know more but the field is exploding so electrical engineering is investing in the field as well and in order to make that relationship work there's a joint laboratory which is joint between electro engineering computer science that has people from both departments in it so because of the limits on the computer science department my offer comes from the electrical engineering department although it's in this joint laboratory so I come out it's an exciting time I'm an experimental sort of on the experimental side of computer science so I'm interested in building things software systems hardware systems things like that and that that effort is being is being rebuilt at Stanford and a person who plays a big role in my life here that I had several senior colleagues Mike Flynn who was the person who hired me and head of the laboratory and forest basket who was on sabbatical when I interviewed but came back a year later and went on to lead a major effort research effort that I became part of Oh again this may not be the way you think about it but broadly what are the questions that everybody are thinking of at the time you know what what needs to be known yeah so this is early on the field you know Apple is just becoming a company the homebrew Computer Club is meeting up at SLAC and on campus it's clear that microprocessors are going to change the way computers get built but it's not exactly clear how the personal computer doesn't really yet exist you know the Apple 2 which comes along a few years later is sort of a very primitive version but much more much more focused on the educational market than general-purpose market a couple years so we're a couple years in several us had gone over to Xerox PARC and seen the alto which really is the prototype that defines the notion of a personal computer as we think of it now and then Xerox makes the decision to donate a set of machines to Stanford and serve all of the universities and that donation is transformative it's transformative because everybody gets a vision of where computing needs to go and while those machines were probably way too expensive to duplicate for every individual person they're a vision of what might be possible in the near future and lots of us begin thinking about how could you use the emerging microprocessor industry and other things to dramatically cost reduce and and make more powerful these these computing costs reduction is a fundamental element in the ladle dispersion yes absolutely absolutely so you're looking for that kind of breakthrough uh-huh around that time so that this so-called VLSI revolution is also happening that Carver mean and Lincoln Way were really key to enabling and that the concept is integrated circuit technology had been held very closely by the by the companies that actually fabricated the integrated circuits so think of it as if they have a technology but they're the only ones who can directly use that technology and the VLSI Revolution was about liberating that technology and letting other people build on of the fundamental integrated circuit economic immersive ace thinking you know I want to hold and profit from it to the more scholarly idea that we share ideas we all will absolutely and multiple many many many more people can contribute if you liberate the use of this technology right separate it really was separate the manufacturing processes from the design process of it and now have lots of design teams not just one or two but many designers yes so at that time farce basket who is sort of my research mentor received funding from from DARPA the Defense Advanced Research Projects Agency to fund a VLSI program and we recruited Jim Clark to come work at Stanford Jim had been a Ivan Sutherlands student and a Dave Evans student at University of Utah and was really interested in using this new VLSI technology to build much more cost-effective real-time 3d graphics so Jim started working on that problem and I started building some tools to help him a programming language which would allow him to implement the code that actually ran on these geometry engines and enabled a higher-level system then to use this capability so I worked on that Jim decided to spin out Silicon Graphics and I spent some time as a consultant there but that company got up and running as Silicon Graphics did very well on its own and we started thinking about what to do next yes this VLSI revolution was still ongoing there was still a lot of thought about it what did we want to work on so I had also been working on some advanced compiler translation technology and so the natural thing to ask was well why don't we design a processor why don't we sign a microprocessor and we were struck
by the fact that the existing micro processors Intel was in the business Motorola was in the business ILOG was in the business Nacional was in the business they had all copied the many computers that had come before them and we we said to ourselves well it's not clear that copying the many computers which were designed for a very different time and a different technology is the right answer let's start with a clean sheet of paper and ask ourselves if we were designing from scratch with the characteristics that microprocessors will have namely trying to get everything on one chip right should they be designed differently and that was our starting point for asking the question that eventually led to us doing the MIPS project and and there weren't that many people asking that question no in fact Dave Patterson was really the other at Berkeley was really the other key group running the risk group he was the other key group asking the question there was a group at IBM but they were really coming from a very different viewpoint that somewhat overlapped ours a group led by John that made some of the same observations that we had made about progress and compiler technology and realigning the hardware and the software boundary because really when we talk about the instruction set it is really think of it as the interface between the hardware and the software so it's a specification that lies between them and up until that time people thought they thought unusual things about where that specification where that levels should be and they thought it should be driven by things other than the one thing that really matters which is where should you draw that specification so that you get the most possible performance at the least cost right and that was how we interpreted the problem as opposed to other abstract principles about it before we get to do you Rica moment yeah it's a process of course it is a process there you will see if the complicated I'm also interested just because of again the history of thinking what is the relationship between the Stanford and Berkeley this turns out to be an extraordinary Center for this kind of thinking how are the two universities are dealing with each other yeah I'd say a friendly competition but also somewhat of an alliance we started actually once our two projects both got going we started doing meetings once a year we'd host the Berkeley people our they'd host us and we exchanged ideas as the as the concepts moved along hmm there were a lot of naysayers in the particularly in industry not so much in the Academy not among the scholarly community more along industry communities who I think I would characterize it as they saw the glass as half-empty they saw what we're clearly academic prototypes they were not commercial project and they saw the glass as half-empty they saw the okay you've done this that's great but when you need to fill the glass the rest of the way you won't be able to do it something I'll break it won't work it won't so that actually strengthened the alliance between the Stanford and Berkeley group because we were on the same side of the equation it's just we knew a little bit about what IBM was doing but IBM was very secretive about
their project and and didn't publish anything and they kept it quiet we had had some interaction with Johncock one of the people who had worked on the IBM project was to had taken a sabbatical and come to Stanford and was working with us and would later on be one of the co-founders at MIPS with me okay again it's not a Eureka moment I know but what's returning so here the first turning point is we start doing simulations of this design we don't yet have silicon we're designing a chip but we don't yet have that back we start doing the simulations and we're getting numbers that are five to ten times the performance of the existing microprocessors that are are we're showing numbers for a single-chip microprocessor there is fat Computers which probably cost a million dollars right then so this is amazing and we we we kind of had this notion well we've simplified the design we've made the instruction set simpler we've made it easier for the hardware to interpret it by lowering it and letting compile or do a little more of the work but we don't have a good scientific explanation of what's happening and that in retrospect that's one of the reasons that there are some naysayers out there it's you discover a phenomena you're hunting in the woods you discover a phenomenal a little bit by chance and serendipity and looking in the right place but you don't really you cannot say concretely why is this the case right it's actually several years later after I started a company that I stumbled onto a paper published by some colleagues at Digital Equipment Corporation and when I read the paper I all of a sudden understood why this phenomenon was occurring and what the fundamental advantages and that was when that was you know when I told Dave we may it was clear what it was and all of a sudden we knew had a we knew how to make it all concrete and clear when you publish the results I mean launched this idea into the world you're still getting grief or have you done worse you know a few years initially we got a lot of grief there was one famous event where Dave and I were on a panel and they had a antagonist who was opposed to our dearest and said this is never gonna make it it's gonna get crushed you know and in the Q&A somebody asks the antagonist here and this was just after I started the company so it's probably middle of 1984 1985 early 85 somebody asked him well Hennessy just got a million dollars to go build a company based on this technology what advice would you give him and he says take the money and go to South America so I still remember that moment we didn't do it and it worked out okay it turned out he was wrong his project got ended up being cancelled actually it was an IBM employee at that time and his project end up not making it and ours helped revolutionize how processors it was a little satisfied yeah that was good it was you know like any startup I mean I was not a I I was not a born entrepreneur I was a real uh we really thought when we published these papers people will understand the strengths the ideas and we'll pick them up but because we didn't have that solid quantitative explanation yet that harmed us in getting people to believe it so they didn't they didn't jump on the idea as fast which is the reason we started the company we started MIPS so because we had to we had to prove the ideas out so again just to underscore this and I
understand it in the lifecycle of a great idea within this framework of computer science sometimes starting a company is not just I want to exploit this and make some money of it it's in the proving of the value of the idea very much so this is very much proving it we didn't really know I mean there was also a team of my co-founders were engineers we didn't know how to general manage him we didn't add it deliver products we didn't know a lot of things right but we we knew and we believed in the technology and I think that was really instrumental I'm not sure what would have happened had we not done that because there was a lot of not vented here syndrome in industry in fact even companies which had engaged in research projects Digital Equipment Corporation which had a major research lab here ended up not pursuing the ideas and so there was a lot of a lot of naysayers and this was key I mean Gordon Bell who had been one of the one of the major engineers and leaders at the jewel Equipment Corporation said to me you're gonna have to start a company this technology is too disruptive it obsoletes too many existing products and people are not going to believe it unless you go out and do it that's good advice and that it turns us over the edge okay later on you're gonna run Stanford at this time you're a humble participant I'm interested in the universities this particularly this university mmm I feel it's this thing relationship to the startup idea - yeah because that's become so important yeah so John Linville who was then chair of electrical engineering encouraged me to do it he he had also done a started a company many years earlier oh I took a leave in my to go start the company in the last year when a year they were gonna make my tenure decision actually which is actually not a bad time to be gone and yeah so I took a leave and I was gone for I was gone on full-time basis for about 18 months and then I came back you know I spent maybe a day a week here nursing my graduate students along so they wouldn't fall off their degree trajectory right um just the slight personal note about your family so is your wife when you're making these radical oh you know she's got her palm trees but how yeah so when we're starting the company she is eight months pregnant with our second child so yeah I had the unit I had my university job and I I had thought I'd go and help get this company started but that I loved the university I long-term I thought I would come back to the University and in fact that turned out to be a stumbling block for raising money because people associated so much of the technology with me even though I had two other co-founders that they were reluctant given that I was had a plan at least that I would go back to the University at some time but you you really thought you would I thought I would I thought I would and I I think what confirmed it not that I didn't I love to being an entrepreneur - it was exciting in its own way it didn't offer I I missed working with students basically fundamentally that was and let's take you back and get you working with students yeah what is that process that's so excite you about doing work I think probably with mostly with graduates most of these graduate students yeah I taught undergraduates a lot and I love teaching undergraduates it's a kind of different experience they colleagues are they graduate students are absolutely colleagues absolutely in the American system they are absolutely colleagues first-name basis lots of interaction we work together we brainstorm together we think through ideas and pursue them as a team so it's a and while there's certainly some mentoring and teaching going on you're helping guiding them as they navigate through a research problem you're helping them develop the skills of relating their research either in a you know talk or in written form but they're doing lots of the innovation and contributing lots of the key ideas as well they get credit they get credit absolutely so one of the things you you do I always put the student names on the paper before so it varies by field but in our field it's very common that people will list the student authors first and then the faculty offers authors after okay so you've in a way pushed a stage of
computer revolution eventually it's going to be accepted by the entireties awhile takes a while like a lot of things okay um so what are your next yeah what's your next research agenda so I'm interested in the interests it given that these microprocessors have reshaped the industry I'm interested now in in whether or not we can use them in parallel processors to build machines which will compete with the very fastest machines in the world but be much more cost effective because they're built out of this low-cost technology as opposed to highly specialized technology and at the time Cray and a few other companies dominate the high end computer industry but they make very bespoke computers as opposed to starting with low-cost micro processors and putting enough together to get that performance so I start working on that problem maybe this isn't particularly relevant to your career but this is also the era where Apple begins to build not only Apple of course the whole the whole structure of this Silicon Valley and so forth but are you aware now of the implications for the area and for industry and I think so you can see so by then by then let's say you know late 80s Silicon Graphics is a major force that's the time that President Clinton and Vice President Gore come to visit the valley for the first time Cisco and the networking business Sun Microsystems there's a shift that has occurred in the in the 70s the the valley is clearly a leader in semiconductor but not in the computing industry you want to go back and visit the leaders in the computer industry you go back and you flood in New York or you fly to Boston and then there's a shift that occurs in the late 80s and you can feel it apples growing Sun is a major force UNIX has become a major force and you can feel all of a sudden that people are coming and then of course late 80's early 90's you've got the founding of Netscape and Yahoo and the beginning to build on the internet industry you can feel chip ting you can absolutely you can feel it you can feel it to try and traffic and the disappearance of the few remaining fruit stands and few trees in the valley disappear you can really sense the sense of the changes this won't be the first time it's discussed the first time you can ask this but very briefly it's an ecosystem or Oh incredible evolving it's not just oh it's incredible ecosystem it's an incredibly strong ecosystem with lots of interaction between the University and Industry with think built a unique ecosystem in in many ecosystems universities and Industry don't always get along too well what one thinks the other lives in an ivory tower are anthe and the other thinks the industry is working on boring problems in fact both they have separate agendas but they're both working on really interesting things and so the ecosystem that get spilled here is one that is much more collaborative and interactive and I think that along with the risk tolerance it's very natural there is though let's talk venture capitalism yeah jumps into this so bencher plays a key role in that and clearly fueling these companies and there's a tolerance for risk you can see people that have started reasonable companies which were not successful come back and rebuild their career afterwards which I think that that acceptance of that and that acceptance of risk was really critical to building the valley okay talking about acceptance of risk in my day in my world which is the museum world if you stopped being a curator and you start doing administration it's like they hold a funeral for you have sold your soul you are no longer doing them yeah yeah okay so tell me about that awful offer and temptation yeah I'm I'm the the so-called frog in the proverbial a bath of hot water that's getting progressively hot I start out doing a few simple things first of all leading this joint laboratory the computer systems laboratory but that's pretty easy that's a 20% time job something like that then finally I'm after turning down department chairmanship opportunities several times it's really my turn as a good citizen yes so I take it on in computer science it's just the time that computer science is moving into the gates building first time the computer science department was in one building since its founding it's give it a year yeah so that's about that's the early 90s okay so and I'm a chair of computer science I like the job I'm I love recruiting young people I love helping the faculty accomplish their goal and discover that I I enjoy doing that so you enjoy not just hands-on direct research but you enjoy supporting others achieve their research and and educational goals so then I met then condi rice asked me to become Dean of the School of Engineering Jim Gibbons who I known for a long time was stepping down after 12 years as Dean and she asked me if I would take on the job so I you don't pause I I thought that is Dean I could manage that and still have a research small research program which probably right small but emphasized small I went from a very large research group with as many as probably more than a dozen PhD students to a few and it happened slowly but it definitely happened the really hard decision for me was when Gerhard Kasper asked me if I would be Provost condi rice had announced she was stepping down and he asked me if I'd be Provost now Provost is a full-time job you might teach from time to time you're not gonna it's going to be very hard to run a research program at a major university of scale and complexity of Stanford and I still remember a key decision point he had me up to the president's house on Friday afternoon and he asked me about it I said can I have the weekend to think about it and it happened that that weekend was Founders Day when we celebrate the gifts the Stanford's made to the University right and the Historical Society had asked condi rice to give the speech as the outgoing Provost okay and she talked about her grandfather who was a black sharecropper in the south and who got the opportunity to go to college by becoming a minister a Presbyterian minister it turns out and how that had transformed the rice family and led to this incredible transformation in their lives her father got his doctorate degree and she went on to get hers and she said that's why I'm dedicated to higher education and it's why it's willing Java's Provost I went in on Monday morning and told Gerhart Kasper that I'd take the Provost job I know you must I'll do that I can't tell her that I did tell her good so you said yes was there ever and this may be a Lobel question rather than a really significant one did you think you had already had your great idea in terms of reason oh that yes so that actually happened around the time I took the Dean's job I felt like this is why I turned down the department job the apartment share job the first few times I felt like my research career had not been well enough established I was elected to the National Academy of Engineering I had an endowed chair so I reached a level of prominence in the field that I felt more comfortable taking on more administrative roles because I don't think I would have done it up in fact I didn't I turn down the department chair job several times because I felt I needed to be a scholar first and have strong scholarly credentials before I could be an academic administrator or a leader right right but you had I had
though it was the right time it was the right time okay I'm actually not going to pursue you into the presidency because I'm sure this has been looked at by many people thought about a lot I want to pull you back to the field and saying now as we come somewhat to the end of our discussion as you look around the field not the state of academia not the state of America but the field that you've contributed so much to what's out there what are the interesting problems where's the interesting work being done right now well we are the we're living with the tremendous discontinuity the breakthrough that's occurred around artificial intelligence okay after after 60 years of I think progress but progress which was probably far below what people had predicted what my colleague John McCarthy had predicted is the person who coined the name artificial intelligence all of a sudden we made a breakthrough that breakthrough came primarily from the deep learning approach inspired by the human brain not exact model of it but inspired by that and it came because two other things
came into play that were critical availability of vast amounts of data to train these neural networks so Big Data the Big Data revolution imagenet lots of different things out there and dramatic increases in computing power were necessary to develop and train these neural networks to act intelligent and that is a discontinuity in the same way that the microprocessor was a discontinuity in the same way that high-level programming languages were a discontinuity here's another discontinuity in the field and it's going we've had a debate at several points with my colleagues at both here at Stanford and at Google whether or not this breakthrough and artificial intelligence would have more impact eventually than the internet and people are on both sides of it I think it will at least be as big as the impact the internet has had on our lives so much of the discourse that I've heard even among specialists but certainly in the general culture of our country is more fear than hope from artificial intelligence why do you what do you think that fear is the first stop I think they're a little afraid that computers will supplant humans and they will supplant some human tasks there's no doubt about it I look forward to the day when we have full self-driving cars everywhere million people are killed on the roads every day we can lower that number dramatically we can't take it to zero but I can probably take it down by a factor of 10,000 or so it's a good point of replacing human there's a good point of replacing humans right I can do that job better clearly there's going to be a lot of disruption in the same way that over a longer period of time the internet created disruption or the Industrial Revolution created disruption a few hundred years ago there'll be disruption some jobs will be disturbed some will be moved elsewhere the challenge we'll have is to ensure that we prepare people for that we provide educational opportunities to enable them to move into new kinds of careers and new opportunities what come Kosovar my last question maybe I'm projecting more than then it's appropriate but another good effect as opposed to oh my god you know people are not going to be necessary anymore is the notion of what can happen within for biology yes absolutely I think the other the whole medical area I think will undergo an enormous revolution as the data becomes available so far we've seen some important small breakthroughs for example looking a photograph of skin and determining whether a lesion is cancerous or not we're a I systems can now be board-certified panels of dermatologists but that's only the beginning because that used some big data that we already had available as we track lots more health data predicting odds of getting cancer and knowing that you should have certain tests or you should avoid certain things which might induce that cancer in an epigenetic effect I think we'll see a lot of progress around that that will improve human health and help us live better happier lives would you where would you send a graduate student today very confident committed to computer analysis a way to solve some of the world's issues where would you send them I mean that is a place but is a field yes I think the the boundary between how you develop the technical approaches to make push this field forward whether it's supervised learning or reinforcement learning but to keep a strong application driver because that's what we really determine how well these techniques work is how strong is the application you need that application driver you need the big data you need the computational resources but when you take that concept when you take an approach an algorithm to the reality of actually having real data and test it
with the real application that's where you see how strong the insight and the algorithms and the invention is that's the last word thank you very much thank you you