The HLF Portraits: Barbara Liskov
This is a modal window.
Das Video konnte nicht geladen werden, da entweder ein Server- oder Netzwerkfehler auftrat oder das Format nicht unterstützt wird.
Formale Metadaten
| Titel |
| |
| Serientitel | ||
| Anzahl der Teile | 66 | |
| Autor | ||
| Mitwirkende | ||
| Lizenz | Keine Open-Access-Lizenz: Es gilt deutsches Urheberrecht. Der Film darf zum eigenen Gebrauch kostenfrei genutzt, aber nicht im Internet bereitgestellt oder an Außenstehende weitergegeben werden. | |
| Identifikatoren | 10.5446/40196 (DOI) | |
| Herausgeber | ||
| Erscheinungsjahr | ||
| Sprache |
Inhaltliche Metadaten
| Fachgebiet | ||
| Genre | ||
| Abstract |
|
The HLF Portraits24 / 66
2
9
14
16
17
18
20
23
25
26
27
29
30
34
35
36
38
40
43
46
50
53
57
58
00:00
Rechter WinkelGrundraumInverser LimesPhysikalismusPhysikalisches SystemMathematikPunktrechnungMAPKategorie <Mathematik>ErwartungswertMultiplikationsoperatorKlasse <Mathematik>Nachbarschaft <Mathematik>RichtungSchaltnetzFamilie <Mathematik>Message-PassingEuler-WinkelFlächeninhaltPeer-to-Peer-NetzBitEinmaleinsVererbungshierarchieGeradeGrenzschichtablösungGradientGefangenendilemmaMereologieProgrammbibliothekComputerspielSpannweite <Stochastik>GlättungKonfiguration <Informatik>CASE <Informatik>ZweiProzess <Informatik>Formation <Mathematik>Kontextbezogenes SystemStandardabweichungMathematikerinBesprechung/Interview
09:29
Klasse <Mathematik>MultiplikationsoperatorProzess <Informatik>ProgrammierungMAPEntscheidungstheorieMathematikMinimalgradKontextbezogenes SystemDatenstrukturNotepad-ComputerHalbleiterspeicherProgrammiergerätBitErwartungswertHardwareDifferenteBefehl <Informatik>AssemblerProfil <Aerodynamik>CompilerHidden-Markov-ModellKalkülGradientLineare AlgebraDifferentialgleichungÜberlagerung <Mathematik>Physikalische TheorieGrundraumPlotterRichtungDiskrete MathematikMultigraphFormale Sprachet-TestGefangenendilemmaVirtuelle MaschineFakultät <Mathematik>ProgrammbibliothekNeuroinformatikInformatikFlächeninhaltQuick-SortPhysikalisches SystemDeterminanteSoftwareentwicklerPunktSoftwaretestPhysikalismusSoftwareGüte der AnpassungRechter WinkelTaskEinsComputerspielProgrammEnergiedichteFamilie <Mathematik>SchnittmengeBesprechung/Interview
18:58
DatenfeldProgrammierungZahlenbereicht-TestProgrammiergerätMathematikQuick-SortStatistische HypotheseNeuroinformatikKlasse <Mathematik>ÜbersetzerbauElement <Gruppentheorie>InformatikSchaltnetzProgrammierspracheProjektive EbeneEntscheidungstheorieNeuronales NetzGüte der AnpassungTranslation <Mathematik>Fakultät <Mathematik>Formale SprachePunktMultiplikationsoperatorTuring-TestSelbst organisierendes SystemSoftwaretestExogene VariableGesetz <Physik>MomentenproblemRechter WinkelGrundraumKanalkapazitätMAPMaßerweiterungMcCarthy, JohnBesprechung/Interview
25:04
Whiteboardt-TestOrtsoperatorStatistische HypotheseMinimalgradHeuristikInformatikProgrammierungMcCarthy, JohnSpieltheorieComputerschachArithmetisches MittelGeradeOrdnung <Mathematik>EinsNeuroinformatikAuswahlaxiomPerspektiveGüte der AnpassungDatenfeldRechter WinkelSoundverarbeitungBesprechung/Interview
28:49
AggregatzustandHardwareOrtsoperatorProjektive EbeneProgrammierungFakultät <Mathematik>SoftwareMereologieProzess <Informatik>Physikalisches SystemSchnittmengeKomplex <Algebra>Metropolitan area networkReelle ZahlMaßerweiterungPunktt-TestSoftwareentwicklerComputersicherheitMAPNetzbetriebssystemInterface <Schaltung>TermTuring-TestNeuroinformatikKlasse <Mathematik>Rechter WinkelATMInformatikerWort <Informatik>FlächeninhaltKontextbezogenes SystemSondierungSemaphorGüte der AnpassungQuick-SortVollständigkeitBAYESEntscheidungstheorieZweiTaskZwischenspracheROM <Informatik>QuaderBetrag <Mathematik>AssoziativgesetzNichtlinearer OperatorBesprechung/Interview
37:09
InformatikGrundsätze ordnungsmäßiger DatenverarbeitungOffice-PaketAbstraktionsebeneLastNichtlinearer OperatorProzess <Informatik>t-TestSoftwareMultiplikationsoperatorBimodulTermMereologieProgrammierungInterface <Schaltung>Fakultät <Mathematik>PunktBitGefangenendilemmaGrenzschichtablösungTaskRechter WinkelKomplex <Algebra>p-BlockModul <Datentyp>ResultanteProjektive EbeneGamecontrollerInverser LimesComputerarchitekturDateiverwaltungProgrammierspracheAlgorithmische ProgrammierspracheObjekt <Kategorie>MatrizenrechnungSchreib-Lese-KopfNotepad-ComputerSelbst organisierendes SystemGruppenoperationNP-hartes ProblemDifferenteMomentenproblemSchnitt <Mathematik>NebenbedingungFlächeninhaltKlasse <Mathematik>CASE <Informatik>ErwartungswertVollständiger VerbandComputerspielImplementierungBasis <Mathematik>Elektronische PublikationProgrammiergerätHardwareHidden-Markov-ModellDatenstrukturWeb-SeiteTechnische InformatikQuick-SortSoftwaretestTuring-TestBesprechung/Interview
45:59
ProgrammierspracheWarteschlangeGruppenoperationObjekt <Kategorie>ProgrammierungDateiverwaltungTermProzess <Informatik>UmwandlungsenthalpieMultiplikationsoperatorFormale SpracheNeuroinformatikCodeImplementierungTypentheoriePunktElektronische PublikationWeb-SeiteRechter WinkelMomentenproblemInverser LimesKlasse <Mathematik>E-MailAbstraktionsebeneZusammenhängender GraphInternetworkingStabt-TestDeskriptive StatistikSubstitutionModul <Datentyp>Turing-TestFokalpunktProjektive EbenePhysikalisches SystemSoftwareBitDifferenteQuick-SortSchreiben <Datenverarbeitung>Mereologiep-BlockBimodulDatenfeldGeheimnisprinzipTaskRichtungUnternehmensarchitekturObjektorientierte ProgrammierspracheTeilmengeGesetz <Physik>WechselsprungWort <Informatik>Besprechung/Interview
54:49
Elektronisches ForumComputeranimation
Transkript: Englisch(automatisch erzeugt)
00:17
To begin at the beginning, where are you as a child? I'm going to say eight. You're eight.
00:26
Where are you? What is your family life like? I'm living in San Francisco. That's where I grew up. I have a younger sister. My father is a lawyer. My mother is a housewife.
00:40
And I'm going to elementary school, public school. Public school. In the city itself? In the city, yeah. In the city. Let's also think a little bit, particularly important in the context of a woman, but in everyone's life. What are your parents' expectations for you? They're definitely expecting me to go to college. They're definitely expecting me to do well in school.
01:05
My father's attitude was that it always had to be all A's. Okay. Not unimportant. There's no sense that the boys in the family might go one direction and the girl the other.
01:22
You were held to the highest standard. There were no boys. There were no boys. Yes, I think that mattered. My brother is nine years younger than me. But when I was really small, there were no boys. And you were the eldest? I was the oldest, which I think is not uncommon. You were the first born? Yeah.
01:41
So really, the expectations are high. Yes. Although I don't think there was, we never talked about careers. You did not? No. It was just, you know, do well. Right. Now, it might be significant or not that your mother is a homemaker.
02:01
Had she gone to university, for example? She had. She was a graduate of UC Berkeley. Okay. And she graduated in, when it was the Depression. She worked for a bit in a daycare center. And then she decided, her father was a pharmacist. And pharmacists were not badly hit by the Depression.
02:23
So she gave up her job because there were much needier people who needed it. So she was just at home. She was a very bright woman. Yes. And in retrospect, I think that it mattered a lot that she never interfered with what I was doing.
02:41
I think sometimes girls get a kind of a negative message from their mothers. Very interesting. And you did that? I didn't get anything about, you should do this or that. I got, you know, just support. Are there books in the home? Oh yeah, lots of books. I read all the time. You read all the time? We went to the library every week. I would get out, you know, as many books as they would let me.
03:05
My mother used to insist that I stop reading and go outside and play. Ah, ah. She was kind of pleased a little bit that you read, too? I don't know. You know, she wasn't a person who necessarily let on to that kind of stuff. She just felt it was good to get fresh air.
03:20
Okay. In California, that's particularly easy. So, I want to decide with you when we begin to see the glimmer of the scientist. Is anything in science particularly attracting you at a younger age? It was more math, I think. It was more math?
03:40
I was just, I just liked it. And I can remember at that age or younger, you know, memorizing the multiplication table. It was just something, you know. And I don't think that anybody was telling me to do stuff like this. I think it was just, it was interesting. I found it easy. Right. Always good for a child.
04:02
So you did that out of pleasure, really. Yeah. To put you in school, really, at either elementary or beyond, what kind of education? California had a pretty good system. At that time it had a good system. I think still there are some good public schools there. Right. And yours was a? It was okay.
04:21
When I was ten or nine, we moved. And then we were in a better, we were in a working class neighborhood when I was little. And then we moved to a more affluent neighborhood. Still in the city? Still in the city. I think that second school was better. Okay. Although by then I was getting close to the end of it.
04:42
So, that was elementary school. Again, because of the limitation of time, I may have to put you in high school suddenly. Okay. What is the curriculum like? Are there mentors potentially there? Is anyone sensing your abilities?
05:01
I sure don't remember anything like that. The only thing I remember like that was that when I was in the fourth grade, I guess I was so far ahead of the kids in math that my teacher asked me to start helping teach the other kids. Really? In the fourth grade? In the fourth grade. But that's the only thing I can remember like this.
05:20
One of the dilemmas very often for women or for minorities and all of those who were not inevitably headed, particularly for a scientific career, math career, is high school counseling, which is suggesting a rather limited range of options. Are you getting counseling in any direction?
05:42
The funniest part of it was, I would say that whatever they were telling me I wasn't paying any attention to. I was just taking the most advanced courses and I took all the math and science courses that were offered. And I don't recall anybody saying I should do this. I mean, it's probably in line with what my father thought.
06:01
But I also think it was interest and I felt at the time that it wasn't something my peers thought I should be doing. So I kind of did it quietly. Quietly. But still, one could argue, perversely. I mean, in spite of whatever the broad message.
06:21
In spite of that message, I was doing it. But the funny story is that at, probably when we were juniors in high school, they had everybody take an exam that was supposed to help you figure out what your career should be. And they came back and told me I should be a landscape architect.
06:41
A landscape architect? And this, I think, was because, I mean, they would never have suggested I become a scientist or a mathematician. I'm sure of that. So this was some combination of some scientific skills and the fact that I was very interested in the outdoors. So a humanist use of your talents. Yeah, something like that.
07:02
Okay, we need to get you to university. How were you deciding that? Well, it was different in those days, or at least where I grew up. There was the wonderful UC system. It was free. And the only thing was that Berkeley was better than the rest of them. But you could get into Berkeley if your GPA was at a certain level. So it was no problem.
07:24
No problem? Yeah. I'm guessing your GPA was at the appropriate level. Yes. It was really different then. The percentage of kids going into college prep was pretty small. And nowadays, I think the competition is huge. I don't think it was as bad then.
07:43
So it was a smooth next step. Yes. Your family had expected you to go to university in any case. Yes. And it wasn't so expensive. And that's where my mother went. Right, and your mother does. Yes. So there you are. I guess you don't have to decide on a major your first year, but how are you picking courses and thinking about...
08:02
So I went in thinking I would be a physics major. Physics major? Berkeley, by the way, has a separate engineering school, and you apply to it separately, and I knew nothing about that. Ah. And if I had realized, I wouldn't have done it, because I really did not see myself as an engineer. Right.
08:20
Anyway, I think physics was because it was the hardest major. But before long, I realized that I was really better at math, and I liked math better, so I switched to mathematics. You're not the only one of the laureates in those categories that started with physics and then switched to math or others.
08:41
So how long did we keep you in physics before the switch happens? Oh, as you said, we don't have to choose at Berkeley. Right. So I had probably already decided this early in the second year or something like that. The courses were the same. It really came down to what did I do after the first two years, because the way it worked then, there were certain physics courses you took.
09:01
I took all of those. And then when you got to be a junior, if you were going to major in physics, the more advanced courses would have gone in one direction, and math went in another direction. So are you noticing that you're doing pretty well in these courses? Especially the math. Especially the math. Yeah. Yeah, but I'm not really thinking about it.
09:21
There were very few girls in these classes. Yes. Were there others at all? In some there were none, in some there might have been one. I was keeping a pretty low profile, and I was the first and the second in the class.
09:41
Was this being remarked on, or you're there so you're there? Nobody's paying much attention. No one's paying much attention. Okay. So when do you do the switch to math? So it was really just more a decision about what courses I'm going to take next year, and I decided to do the math. So it was still within the context formally of a physics major?
10:05
Well, I don't even remember when you declare a major. Okay. But I think that by the time you would have started to make the decisions that actually mattered. So the first two years I was prepared to major in either one, and then I went off and I just took many more courses in math.
10:22
Right. At this comparatively early stage, although in math one's abilities come earlier. Yeah. But are you finding a particular direction in math that's fascinating you, or are you just hungrily consuming? Yeah, I'm just consuming. I think I had realized I like discrete math better than continuous.
10:47
Define for me the difference between discrete math. Well, things like set theory, linear algebra, as opposed to calculus, differential equations. Got it. And, you know, now I can see that's a lot more like computer science.
11:02
Ah, but of course then you're just following your interests. Yes. You graduate with probably some distinction? I can't remember. I graduated, I actually applied to grad school in math, and I was admitted at Berkeley.
11:23
But I decided that I shouldn't go on right then because I wasn't dedicated enough. Well, now you'll know I've done some background work because you applied to Princeton, but Princeton was admitting women. Didn't admit women. This is so inconceivable now that I want to stay with that a bit.
11:42
Did they announce this? Did nobody advise you not to apply? Were you aware of this restriction? Not until I got a postcard back from them. Were they stated that they did not accept women? Yes. In math? Yes. Unfortunately I didn't keep this postcard. So, you know, and memory is not that reliable. So this is my memory of it.
12:03
Yes. Yeah. That would belong in a museum. If I had that. If you had that. I was surprised. I mean, Princeton at that time did not have women undergraduates. So I knew I couldn't have gone there as an undergraduate, but I thought at the graduate level it would be different.
12:21
What is the year now? 1961. 1961. Because I know that in the 50s, Harvard Med actually had female faculty, but women could not be admitted as students. I mean, it was a very bizarre time for this.
12:42
This is undergraduates you're talking about? No, no. Graduate students. Okay, so it was. Yeah. Not unlike your situation. So it's curious. They had distinguished faculty, but women were not allowed in the medical school. I have talked to people who were undergraduates at Radcliffe, and I think it was good that I didn't go there, because they have stories about how
13:06
they could take the courses with the men, but they had to take their exams separately, and they couldn't go to the library at this and that hour. And it really seems like a century ago, but it's not really so long ago. Not so long. All right, so Berkeley admits you, but you decided not to go. Why?
13:24
I felt I wasn't dedicated enough. I didn't really want to go on to that next level. And I think under the covers, though I didn't really bring this out, I realized I wasn't as good as I needed to be, because, you know, math is a very, there are people who are just obviously going to be the
13:44
math geniuses, and if you're not quite at that level, even if you're very good, you're not going to get there. And I think I knew this, although I didn't articulate it. I thought of it more as, to go to graduate school, you had better really commit yourself, and I wasn't ready to do that. But you wanted to begin a career somehow related to your education.
14:03
I wanted to use my education to get a job that paid a decent wage. Fair enough. So where do you wind up? Well, I decided to go to Boston because my father grew up in this area, and I'd never experienced it, and I had a friend who graduated from Stanford who was interested in doing this, so we decided to do it together.
14:23
And she had majored in biology, and she got a job as a research assistant in a lab at Harvard. And I came here without a job, and I stayed with my aunt and started applying for jobs. My aunt told me that I would never get a job, you know, I'd only be able to get a job as a secretary, and the wages would be very low and so forth.
14:48
Anyway, I was applying for a job based on my math, and I did have quite a few job interviews. And the jobs that were offered me were not interesting, they were, you know, plot graphs,
15:01
very low level math stuff, which really is not surprising given what math is really like. But I got a job offer as a programmer from the MITRE Corporation. Which begins the rest of your life in a way. It does. And that sounded a lot more interesting. Yes. And tell me a little bit about what the job was like actually.
15:21
The job was actually very funny in retrospect. Well, first of all, computer science was really new then, and there weren't people coming out with undergraduate degrees. So they were hiring anybody they thought might be able to do it. And so they were taking a chance on me just like they took a chance on a lot of other people.
15:41
Many of whom were women. Many of whom were women? Yes, not all of whom were in math, some of them were in English or something like that. You know, either you can think this way or you can't. Right. And so it sort of all, you know, works itself out. I get that, I'm guessing you're doing pretty well at this job. So they handed me, they handed me on my first day of Fortran manual, and they said, write a program to do this.
16:05
So I'm entirely self-taught, but yes, I was really good at it. And I realized, you know, it was what I really liked. So it's a solution to the math dilemma. Yeah. In a way, you're finding the best outlet for your intellectual energy.
16:20
Yes. I think you're there only a year? I decided after a year that, I was living in Cambridge, and wouldn't it be nice, I saw an ad at Harvard in the Harvard Computation Lab. They were looking for a programmer, so I thought, well, that'd be much more convenient.
16:40
So convenient. Yes. You made the difference in the jobs. It wasn't that the job was better. It was easier to get to. Much easier to get to. And they actually offered me a higher salary. And the first salary confounded my aunt's expectations of the second one. And at that point, MITRE tried to, you know, keep me.
17:03
But since I was madder, I really preferred to work in Cambridge. There was no way that was going to happen. Fair enough. So what are the tasks that you are assigned at Harvard? Harvard turned out to be a really good change, because at MITRE I had been working in FORTRAN,
17:21
so I'd learned how to use a higher level language. At MITRE, I was maintaining their huge machine program, a program written in assembler. So then I began to see what the underpinnings of a computer programmer like, what the compiler does, and what really happens when the hardware runs. I also got to see what a really big program looked like.
17:44
And I think it probably was the beginning of a lot of the stuff I did later, where I was thinking about how to structure large software systems. So the move of convenience turned out to be a determinant in your intellectual development. It did, because it really broadened what I had learned at MITRE in this new direction.
18:04
So how did we get you to graduate school finally? Well, it must have been in that fall, because you have to apply in the fall early, that I decided that I was learning really fast, but it was all self-taught, and I might learn a lot faster if I went back to school.
18:23
Plus I think I was ready to go back to California, because I'd been away for two years, and my family was there. Right, of course. So I applied to Stanford and Harvard. Stanford and Harvard. I didn't apply to MIT, because I thought of MIT as an engineering school
18:43
where I wouldn't want to be. In which schools did you apply? Just Harvard and Stanford. No, but I don't mean universities. I mean which direction, what school of what? So Stanford had a program in computer science.
19:03
It was a sort of combination between engineering and math, and so I applied to that, and I applied to Harvard. I don't even remember, but since I was working there, it would have been in what they call applied math. Okay. This is obviously the beginning days of the capacity to choose a computer career.
19:22
So this is the way that each university is shaping its development, developing program. So Stanford accepted you, and perhaps it's romantic to say, but that must have been an extremely exciting point to wind up at Stanford. I guess. I don't remember that.
19:41
It fit what I wanted to do, which was to move back to California. You know, I wasn't getting guidance, and I didn't understand that Stanford is one of the three places in the country that you really wanted to go. You didn't. I did not know that. Good fortune. Yes. Which actually affected both what you did at MITRE and Harvard too.
20:03
There's some good fortune, luck, but there's also your response to the luck. That's right. I think that everybody has stories like this. So there you are at Stanford. You don't know that you're a place that is beginning so many things. You're expecting a Ph.D.
20:23
to finish the course in what? What would that be? Computer science. It would be called computer science. Well, you know, again, this is a sort of a formal definition that I knew nothing about. Right. But I was in a program that was really computer science, and so whatever they called it, it would have been computer science. As it turned out, they turned it into a department a couple of years after I was there.
20:42
So again, a good moment. Yes. A good moment. I think there's a particular professor who influences your time there. John McCarthy was my advisor. He was a Turing Award winner. And my recollection is that I went to Stanford without any financial support.
21:03
I wasn't really worried about this because I had been saving all the money I had made, and I just didn't think about it. But my recollection is that I met him walking up the steps on the first day at Stanford, and I asked him if he would support me. And he said yes. Now, it's highly unlikely that this is the whole story or even true.
21:24
But it's essentially true. I mean, it may not have happened only for that reason. My guess is they admitted me because I'd been working at Harvard in the quote language translation project, and so they thought of me as being an artificial intelligence person, even though I was just a programmer.
21:41
And so maybe they had already thought I might end up working with McCarthy. He had a lot of research money, so he was looking for students. But it was kind of happenstance. Maybe, you know, if I met somebody else on the first day, I might have grown. But happenstance is important. So there you are. You're in his lab, essentially.
22:01
Yes. What are the kinds of problems that you're addressing, and to what extent will it affect your dissertation? Now, that's a problem I have trouble answering. Stanford does not have a master's thesis, even now, and didn't have it then.
22:20
So there was no notion of a first project that you completed and wrote up. We have that at MIT. I think it's actually helpful because it can be small, and it gets you sort of started. So I was mostly, you know, I don't know. I really can't remember. Fair enough, fair enough.
22:42
But you have tasks. I'm working on classes. I seem to have been writing little papers about this and that. I don't know why. In retrospect, clearly that was a little unusual, but it didn't seem unusual to me. At one time, Nicholas Veert, who was also on the faculty at that point,
23:04
he's another Turing Award winner who is in programming languages, tried to get me to switch over to programming languages because I was interested in things that were going on in compiler construction. And later on you will be very interested in these kinds of things. Yeah, I never really went into compilers, but programming languages for sure.
23:23
But I decided, that was probably third year maybe, and it just seemed better to stick in AI because I probably could get done and finish sooner. Right. Yeah. A good kind of decision for a career.
23:40
Are you looking around and seeing any other women? There's a woman the following year. This is Sue Graham, who became a professor at UC Berkeley. There was another woman in that class. In my class I was just me, but it was very small. Maybe there were certainly no more than ten. Some small number of students.
24:01
They never had a kind of a class organization where you said, oh here's your class, but it was really small. And Raj Reddy, who was another Turing Award winner, we actually worked, we didn't work together, but we both worked for McCarthy, so we knew each other quite well. And anyway, so I was the only girl in my class.
24:23
In the next year, when Sue came, there was a second girl who ended up not staying. So there were, you know, one or two. One or two. In the broader question of collegiality among the graduate students, is there a lot of interchange going on? I felt they were very collegial.
24:41
And we used to take classes and we'd get together in the computer room at night and run art programs. I never sensed any sort of, I would say it was much more collegial than anything I ever experienced when I was a math major at Berkeley. Oh, interesting. So you are, is there a sense of excitement about an emerging field
25:04
that everybody is the ground level of or people just... Well I would say that definitely the professors probably sensed this. I think the students are more, I mean they chose the career for a reason, but I don't think we understood what it meant.
25:21
I certainly didn't understand what it meant to have lucked out and gotten into a field that was just emerging. Yeah, fair enough. What is your dissertation topic? It was a program to play chess end games. Describe what the inquiry was.
25:41
The inquiry was whether, well to give you a background, you need to understand that computers were not very powerful in those days. They were slow and chess is basically a search game. So you start with a particular position on the board
26:02
and you ask the question, if I make this move, what will the counter move say? You're thinking about what's the end position. And today the computers can search very deeply. Then they could hardly search at all. So in any way still you have to prune the search because it just gets gigantic.
26:25
But then you used to prune it very aggressively. So then the question was, well what's the most effective way to prune it so that you don't throw away the good paths, but keep the ones that are worth pursuing. And so John McCarthy had a hypothesis that for the end games,
26:42
you could use this notion of better and worse. So you had a way of analyzing a position that would tell you whether it was better than where you are and maybe better than some other choices that you had to make. And so he wanted to see whether that was effective.
27:01
And he thought I would be a good person to work on that because I didn't play chess. And therefore I could go read the books and I could see, which I think maybe this was correct, I could see in them the ways they were expressing, they were explaining people what heuristics they might think about using. I could think about it more as a program than like something I'm really familiar with.
27:23
So I worked it out for a few end games. It worked for those end games. I never thought it was really very important, but it was again a means to an end. By then I had already decided I didn't want to be an AI. You had already. I had already decided that, yep.
27:42
That was also along the lines of, you know, I would have probably preferred to have worked with Veer, but I decided it was better to get done. Yeah, yeah. There's a practical streak in you. Apparently, I mean, we're so into this. So, you know, much is now made of the fact that you're one of the few women,
28:02
first certainly women, to get a PhD in computer sciences. Again, I keep asking this, and probably the truth is it wasn't as big a deal then, but tell me, was much made of this? No. I didn't think about it at all.
28:21
You know, I was just one of the students, and I think Raj got the first degree, and I got the second one, or maybe Bill McKeeman got the first one. I mean, we were really early, but I wasn't thinking about it from that perspective at all. Good. I mean, you're not bearing this burden of the first. You're just doing the work. I'm just doing my, yeah, I'm just doing my thing, right? You're doing your thing.
28:41
Okay, now, again, at the cusp of a post-PhD career, very often, certainly later in computer science, but maybe at this point or not, there's the question of do I go into academia? Do I do industry? How are you thinking about the next step?
29:01
I was thinking about academia. I, again, didn't have any guidance. I didn't even know to ask for guidance, but I was looking around. I did apply to some places, and the only academic job position that I got was at,
29:25
I can't remember which one it was. It was some state school in California on the other side of the bay, like Milpitas or something like that, and I knew that wouldn't be a good decision. Such an area. Well, from an academic position, yes. From an academic position. You know, it would have been good teaching, but research-wise, nothing,
29:42
and I could have stayed and worked at SRI, and that would have been a good research position, but I had met my husband by then, and I wanted to go back to the Boston area. You met him within, at Stanford? I met him while I was a student. While you were a student, but he wasn't at Stanford. No, he was out on business. Okay.
30:01
And so I applied to MIT, and they offered me a research position. Okay. And I applied to MITRE, and MITRE offered me, and at MIT, it would have been a, I was applying as a faculty member, so the fact that they weren't willing to give me a faculty position
30:21
meant it was some sort of second-class thing. Yes. Again, I'm going to inevitably ask, do you think it was affected by the fact that you were a woman? Oh, absolutely. Absolutely. Absolutely. I mean, if I hadn't been a woman, maybe they wouldn't have been interested, but I think actually there was an old boys' network at work. I didn't know that. Right.
30:40
And so they were, you know, hiring the students of their friends, and these would always be young males at that stage. Okay. Well, MITRE doesn't have this problem. Well, MITRE knew me, you know? MITRE knew me, so they were interested to hire me, and that was a real research position. It was in systems, which is what I wanted to switch to,
31:02
or it could be whatever I wanted, and... Really, come to us and decide what you want to do? Well, they have, you know, they have various projects they're working on. The way it works at MITRE when you're a young researcher is they have various things, funding for various projects, they've lined up and they give you something, so they gave you something.
31:20
You're very interested in the task at hand? Very. This is suiting you? This is, yeah, fascinating. And I think of it now, and somebody asked me, in one of these, I don't know, someone I was talking to, didn't you find it kind of scary that they had asked you to do this stuff which really wasn't like anything you'd looked at before? And no, I didn't.
31:42
I found it fascinating. It was like a box of candy they presented me with, and they gave me complete freedom. I mean, I just got to... And probably a decent salary. Oh, yeah. For those days. For those ages. Yeah, right. So I think it's in this context that you create with associates
32:02
or maybe entirely, and you're on the Venus computer. Yes, the first part of the project I did with an associate, his name was Bob Curtis. He didn't have a PhD, he's a very bright man. And we did work together on that. And then the second part I did entirely on my own. That was the software part that worked on top of the hardware.
32:20
Can you describe what it was doing, you know, as a project? The end game, if you will? It was a project to invest something called micro-programming. Micro-programming was invented by Morris Wilkes, who was one of the early Turing Award winners. And the idea was
32:40
to have an intermediate language above the hardware, but below the instruction set that people use. So you would design the instruction set, interpret it in terms of this low-level stuff, and then it would run on the hardware. And this made the job of designing the hardware easier because
33:00
they had a simpler interface to design to. And so I was working with a particular piece of hardware, the Interdata 4, I think. And my job was to invent an instruction set and implement it using micro-programming. The micro-program went into a read-only memory, very small.
33:21
And then you could run anything people wrote in this instruction set and would interpret it using the hardware instructions. So the first job was to design that instruction set. And you couldn't do very many instructions, but a lot more than what you had available at the hardware level. And one of the things that was interesting was we decided to put semaphores
33:42
into the hardware, and that was an idea that came from Dijkstra, another Turing Award winner. One of the things that's been wonderful in my career is I got to know all these people. Yes, of course. MITRE is still a corporation. It's got profit motives, it's got all sorts of things.
34:00
It's a non-profit. It works for the government. I see. It still works more or less in the same mode it did then. The government puts out requests for proposals, MITRE bids on them, they have certain specialties. Oddly enough, my son works for MITRE. And he's a computer scientist, but he's insecurity.
34:23
So nobody is worried about monetizing anything you do? Oh, I think people weren't worried about monetizing. The government wanted to know, was this a viable thing to do? It was really research. How long were you at MITRE? I was there four years,
34:41
because after the Venus project, then I went on to do the program methodology stuff, which was really the start of the ideas that led to the Turing award. I mean, this is the way MITRE worked then, and I think to some extent it still works now.
35:01
I was finishing up the Venus project, and they asked me to look at this problem of the software crisis. So this was another place where somebody had bid on a proposal from the government and so they were looking for somebody to do the research. so I started to look into this.
35:21
And the software crisis had to do with the fact that they didn't know how to build software. And the programs they were building were not working correctly, and they would have to sometimes abandon the entire project. And even though it was early days, some of these were pretty complicated programs, ballistic missile guidance and stuff like that. And so my job was
35:41
to start to look into this. This was probably in some sense much less directed word search than the first project was. There it was, you know, find that instruction set and show that that instruction set is useful. But here it's more, here's a huge problem, see what you can make of it.
36:01
good for intellectual development. Much more like being a professor. More like being a professor, exactly. What do you make of it? Oh, it was fascinating. I mean, so I read all the papers, nobody had any answers. You know, there were some people that had suggestions here and there. And so I was thinking
36:20
about it, and along the way I realized that in designing the software for the Venus operating system, I had actually put a programming methodology in place. Not because I was thinking about programming methodology, but because I was worried about how are we going to manage the complexity of the software. So I was already in some sense thinking
36:40
about the topic of programming methodology without realizing it. Right, you were prepared more than you knew. Yes, well maybe that's always true also. Anyway, so I wrote that up. And in the meantime, I had written a paper on Venus and submitted it to SOSP, the systems conference,
37:03
and it was accepted and it was an award paper. And when I went out there, and this would be in the fall of 1970, it could have been, it was probably the fall of 1971. And so Corby, Professor Corbató,
37:23
another Turing Award winner, was in the audience, and Jerry Saltzer, a professor at MIT, was the head of my session, and they were then actively looking for women. So, and so, and there probably were, probably Jack Demas was there too, so there was a group of people from MIT, and again,
37:41
I'm very naive, so I don't really know what's going on. But as a result of this, I was invited to apply. To MIT. Which is not a bad invitation. No, and I was actually invited to apply to Berkeley too. Really? And I didn't even bother, I didn't go because I told them
38:01
there was no way I was going to move back to California because my husband's job wasn't moveable. You were now a Bostonian. Yeah, at least for the short page. And here it is, you know, almost 50 years later. So, you had sat. So I had to apply.
38:21
I gave a talk, you know, honestly, it must have been on Venus, but I don't remember what it was about. I remember a little bit of the interview process, anyway. You seem like a nervous person, so you're not, at this point in my guessing, so wracked about, oh my god, will I do well, will they take
38:40
me or not? Hmm. You know, I don't know. I don't remember. I had a good job. You didn't actually need this. I didn't need that, but it was something I was interested in. So I was just exploited to see what would happen, but when they offered me the job, I took it instantly, because
39:00
it was definitely, you know, it had been what I thought I would want to do. Another thing that happened while I was at MITRE was I changed my expectation about what my life was going to be like, because I had thought that, and this was not uncommon for women at that time, that I would work until, you
39:20
know, I had children. It used to be you would work until you got married and then they'd want you to leave, you know. So now you've had a few more years. Yeah, so you work until you have children, and I used to think about, you know, this was something I was doing, and then after that I would do something else, but I realized when I was at MITRE that this was not true, that I really wanted to do this stuff, and I wouldn't
39:40
stop doing it. Very important. You're at MIT. What is your rank? I'm an assistant professor. You're an assistant professor, and what are you either tasked to do, or what do you choose now to do with this new opportunity? Well, so there's the task and the choosing. The task is teaching,
40:00
and MIT is to this day a very strong teaching school, and all faculty are expected to teach. The faculty load is one course a semester, and there wasn't actually a computer science department there. There wasn't even electrical engineering and computer science.
40:21
I was hired in double E. Really? And the course they gave me to teach was a course about computer architecture. These are team taught, so I was teaching a recitation, and it involved quite a bit of hardware, which I knew nothing about. So that was a real
40:40
scramble, and not only that, but I don't think the students were very happy to see a woman. There were definitely students in the class who were trying to test me, and given that I was two weeks ahead of them, and maybe not two weeks ahead of those, I had to learn how to manage that. So that was a hard
41:01
first semester. Fortunately, I did teach that course several times, because once you do it once, it's much easier to do it the second time. And you did survive it. And I did learn how to manage the students, and you know, yes. The interesting thing is I got this from the students. I didn't get this from my colleagues.
41:20
And but then on the other side, I have to find my research. And the research was, the way it works in the electrical engineering department, the ECS, is they have these labs. And the labs kind of cut, it's like a matrix organization. They cut the departmental structure. And so I was in the lab for
41:41
computer science. And no, actually there was just Project MAC in those days, because ECS didn't exist yet. And they had decided that I should be in AI. So they stuck me up on AI. Right. So they stuck me up on the AI floor, and they even asked me to go work with somebody or other.
42:01
And I wasn't having any of that. But that was pretty miserable. So I had to figure out how to sort of get out of this dilemma. Meanwhile, I'm thinking about program methodology, because that's what I really want to work on. Right. So in a way, I had no trouble with the second part, because I already knew the research I wanted to do. But I had to kind of manipulate this. And Jack Dennis,
42:21
who was a professor there, was very helpful to me. He was really a mentor, and he helped me move my office to a different floor, and sort of encouraged me to look into the research I was doing. So he was helpful. And you got it right, I mean, in terms of your interests. Yes, that it was the right place to be, and he was very interested
42:41
in the work I was doing, and you know, so that was very helpful. Is it the CLU programming? Well, the first year was actually writing the paper on data abstraction, which was the basis for CLU, and all the stuff that happened after that. And then starting in the second year, it was designing
43:01
CLU. And Jack used to come to those design meetings. And I mean, in the end, we didn't really work together, but he was just really good at making me feel supported. So then a bit of luck, this collegial relationship. What are the
43:21
objectives, what are you setting out to achieve, or what is the problem at hand? So the problem is this very abstract problem of how do you organize software so that in the end you can control the complexity and make sense out of it. And the solution is modularity. So you have to
43:40
break the software up into modules that where they have an interface that describes what they do. And then inside might be a really complicated implementation, which might make use of other modules. So it's a hierarchical and you use ideas like this in lots of
44:00
areas, certainly in engineering. Sometimes the modularity comes out of physical constraints. Here though it's entirely intellectual. It's the only by having these simple independent blocks that you can reason about independently can you really keep control of stuff. And that was what people were struggling to come up with. At that time this kind of stuff was not understood.
44:22
And data abstraction is a new kind of modularity. So, and it was a huge step forward to come up with that idea. The only modules they had before that were procedures like, you know how to sort. That could be something separate from all the rest of what the program was doing. And data abstractions do more because they have, it's like a file
44:42
system is a data abstraction. There's many operations on files. There's a huge amount of complexity down below to get it all implemented. So that kind of thing. Maybe this is too melodramatic a way to ask but is there a eureka moment at this point? Oh yeah. There was although I don't remember it specifically but there was a moment
45:02
when all of a sudden I saw that there was such a thing as data abstraction. And prior to that I had been working just with I mean I had just been thinking about this. All these papers, my own ideas about program methodology and trying to find a way that would make these useful in programs
45:21
and they didn't remain so abstract. And when I thought of data abstraction I realized that this was something you could put into a programming language. This is something programmers would understand because they already understood about procedures. So abstracting from how do you do something to accomplish a particular task was already something in their repertoire. It was just this was
45:42
a different kind of task. So yes I knew when I thought of that that this was a really good idea. And it was. And it was. And it was. And it was terrific. Again because it's such a rich career and such a limited time I want to
46:01
jump into the process of creating the Argus language. Ok, so I worked on Clue for most of the 70s. And with a group of students, you know of course you don't ever work alone when you're in an academic career. And when it got finished around about
46:20
1978 I was looking around for what to do next and one thing I was thinking about was continuing the work on Clue and maybe, but I thought it would just be pretty incremental to keep working in programming languages at that point. And then I discovered a paper of Bob Kontz who's another internet
46:42
give me a Turing award winner and I discovered the internet. Which I had not been paying attention to although I had been using email already for several years. And I discovered in reading his paper by Bob that there was a kind of a dream of writing distributed programs
47:01
where there were components on different computers connected by a network but people didn't really know how to do that. So I thought, you know, here's a great research project. And it was. And it was. So Argus was the first part of that project and I designed a programming language. How do you organize
47:20
that task with colleagues and so forth? Do you say this is my next interest? Do you then find just the process of creating a team? So I always had students who wanted to work with me.
47:43
You have to get money. But money was easy to come by in those days because we got a big block grant from DARPA and all I had to do was write a couple of pages and of course DARPA was where this came from in the first place so they were very interested in this. I don't think Bob was there
48:02
anymore at that point but I can't really remember. So anyway, you know, so I had money and I had students in my group who were looking for things to do and it was a programming language project so we started working on a programming language. I didn't do much of anything except you know, sort of keep on
48:20
going and explaining new ideas to people. And I used to write, I would write NSF proposals also there. They don't bring in the kind of money that the DARPA grants do but they, you do explain your ideas so they're good for that. Right. I assume part of the explanation is
48:40
some sense or guess at the implications of other work. That's right. So why is it important to do this? How are you going to go about it? So I wrote something but I don't remember what I wrote. At what point do you see success in this enterprise?
49:02
So well we had a goal. We were going to design and implement Argus and get it running on multiple computers. And I had a group of students working with me on that and I even had a staff member.
49:20
You know so by then there was a lot of programming involved and I had to give up programming because there's a limited amount of time you have and programming is very time consuming. Yes. So I was doing less and less programming myself. You would ask for somebody to work on something. Well you would describe you know what needs to be done and I would certainly be
49:41
involved in design details but the actual writing and debugging of the code. Yes. Yes. And somehow I got together enough equipment so we could do these experiments but that again was not difficult at that time because there was, we had a lot of money. Yeah. Yeah. And you know it wasn't necessarily
50:01
It was kind of a golden age in terms of Yeah I would say problems and well I guess there are always interesting problems but support and maybe even acceleration of insight. It certainly was a very good time to be doing research. Right.
50:20
I know that there's such a thing as the Liskov Substitution Principle but I don't understand that. Okay so that was later. Okay. What happened with Argus though was that it sort of pushed me into distributed systems which has been a major focus of my work ever since then and
50:41
I went off in many different directions based on that. That was just another interesting field. The Liskov Substitution Principle which has turned out to be extremely influential came about not because I had a research project looking into this it was because I was asked to give a talk at the
51:00
Uppsala the object oriented programming I forget what the acronym means but anyways the main object oriented programming conference I think it might have been the second keynote so it was the second year they ran the conference and so I said okay and I at that point I decided well I probably better find out what's going on in object oriented
51:21
programming so I was I was doing the data abstraction part and then in California Alan Kay's group another and and other people in Silicon Valley were interested in what's called object oriented programming which has this notion of their objects and their hierarchical and you can
51:41
implement a new kind of object by taking the code of an old kind of object and just adding to it and changing it a little bit and those ideas came from Simula 67 which was Olio and Kristin so they were working away
52:01
on object oriented programming and I was working on data abstraction where the focus was much more on encapsulation modules that had hidden implementations on the inside and the outside was described by a specification and I was interested in how do you prove the correctness of the code and how do you reason about stuff
52:21
and so I hadn't been reading the literature that was more the object oriented literature and I decided to read some of it and I discovered that they didn't really have this strict notion of modularity they had more of an idea of just developing code based on other code and they were kind of struggling for
52:41
some notion that was like this because they were talking about what they call types and subtypes and by which they meant you know a stack is a type a file system is a file is a type there are many things like that and they were interested in you know what's a subtype of a of a queue like thing
53:02
you know there's a stack there's a specific FIFO queue and so forth and so they were clearly looking for these definitions but they didn't have it right and they were very confused and in fact their description of the behavior of classes was often implementation driven it's like this but you change
53:20
this implementation and so hopefully that's not but so I'm trying to remember where I was right so I read these papers and I realized they were looking for something they didn't have and it was clear to me what it needed to be it was just a sort of
53:41
common sense idea and so I those are often the toughest to come up with well you know maybe coming from my background in data abstraction I was able to see it and because they weren't thinking in those terms and thinking more of these implementation terms and so I gave the keynote at Uppsala and then later I wrote up a
54:02
paper describing it which was never actually published except in I mean later 10 years later we published a paper but at that time no and but it just the ideas took out like wildfire because they were really looking for it and I discovered in the 90s when somebody wrote me an email they asked is this the correct definition
54:22
of the Liskov substitution principle and I didn't even know there was such a thing the name was invented it was on the internet everybody was talking about it and then later working with Jeanette Wing we wrote some papers to come out with precise definitions of what it was well I think this interview should end
54:41
at that moment of triumph oh okay thank you very much oh you're welcome
Empfehlungen
Serie mit 66 Medien