Computer Oral History Collection, 1969-1973, 1977
39
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
HOPPER:
It is from that point of view. It was solely Aiken not (Harvey?). Aiken sold the concept to
IBM who built it and then he told them to give it to (Harvey?). And he gave them all his
rights and patents in order to get it.
TROPP:
You may not want to talk about this, but this was a question that you raised. Maybe you
can phrase it better Beth about the role and the clash and the whole difficulties that later
developed between Aiken and IBM.
HOPPER:
IBM never gave the credit they should have to Aiken. They gave credit to their own
engineers but they did not give it to Aiken.
LUEBBERT:
We were reading IBM's…(voice fades out).
HOPPER:
When they came down to take the pictures when Watson visited us, they took the picture
of Watson and the IBM enlisted men, but not the officers that ran the machine. Not the,
just the pictures of the ex-IBMers.
TROPP:
Had that kind of clash occurred earlier or was it…?
HOPPER:
It had certain begun to occur by the time I got there. IBM was about to grab the whole
credit for it because their engineers had built it.
LUEBBERT:
Was that because Aiken was down at (Dalvin ?). I guess it was. While it was being built,
that they wanted credit for it?
HOPPER:
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
40
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
They wanted credit for all the ideas and it was true that the multiplier had been invented
ahead of time by some of the IBM engineers. They built the multiplier which was later
incorporated, the concept was incorporated in the Mark I.
The overall concept and in particular the secrets mechanism and the interpolator, were
purely Aiken's. And IBM tried to take prime credit for all of it. They couldn't conceive of
anybody but IBM building it. They built that horrible mess known as the selector
sequence control (calculator ?), which is one of the worst ( ?) that anybody has
built.
TROPP:
One of the other things that…
HOPPER:
Watson did not give the credit to Aiken that he merited.
TROPP:
That's the…
HOPPER:
That's the scoop.
TROPP:
Of course, there's another view that I…
HOPPER:
Plus the fact that Aiken wouldn't cow tail to Watson when Watson …(voice fades out).
I've had my own experience with IBM in '49 when I started job hunting. When I, when
my three year contract with Harvard terminated. When I was job hunting, I went to him
for an interview in New York at IBM. They gave me a three or four paged thing about
this long on yellow paper to fill out.
And I had gotten to one and a half pages when they started singing songs to Watson in
the next room and I handed them back the blank pages and said, this is no place for me.
(LAUGHTER).
I have a record of those IBM songs if you want to hear it sometime. (Which I?) liked to
give them away a few years ago.
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
41
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
TROPP:
We just got a copy of that. Yes, in fact…
HOPPER:
And they really did sing them. That's the thing that soured me and I never went near IBM
again.
TROPP:
We also have a copy of the 1934 song book.
HOPPER:
I couldn't live in that kind of an environment.
TROPP:
Well one of the things that struck me was another point of view in the opposite direction,
when I look at the dedication records for the Mark I. It's essentially a Harvard
announcement which tends to exclude everybody else.
HOPPER:
Well that was Harvard and typically so and that annoyed Watson too. But that's Harvard.
TROPP:
Yes, I was going to say a Harvard inauguration of the Mark I. Well going back to the
Mark I, I want to go back to the question I asked you earlier in terms of its major
contributions and its ultimate impact in the subsequent developments and some of the
ideas that got lost.
HOPPER:
Its major impact was through the people who worked with it. We carried it away with us.
TROPP:
That's one of the things that struck me, that was the…
HOPPER:
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
42
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
The basic concept you see, for compilers, stems from Harvard which I took with me. The
ideas which I was generating on how to write programs easier at Harvard I finally
brought to a fruition ( ?).
It was the people that, who were here and went somewhere else who took it. Because we
were all scattered. ( ?) and everything was scattered and we took them with us
and they grew everywhere. So that the impact is far greater than people realize.
TROPP:
Well that was the…
HOPPER:
That we took them with us.
TROPP:
…the conclusion that I had come to sort of tentatively that Harvard was the training
ground for the future. It performed a major function.
HOPPER:
Very much so. And the next big one was the UNIVAC I believe it or not.
TROPP:
Oh yes.
HOPPER:
The UNIVAC people are all over the industry far more than IBM. If you look at that
history of software outline that I made of the development of software, you will find the
origin of most of those concepts came from UNIVAC. And that's Betty Holberton and
so…(voice fades out).
And you see, for instance, Joe Harrison, you know came to UNIVAC before he vanished
with APL.
TROPP:
There are two interesting parallel developments going on …
HOPPER:
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
43
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
Both of them came to UNIVAC. Katz came to UNIVAC. Many of those Harvard people
came to UNIVAC from what I heard.
TROPP:
At the same time Beth, there's a parallel development going on in the West Coast of a
similar kind of thing.
HOPPER:
It spread through people. Very much so. Not through the colleges.
TROPP:
So, in looking at the developments historically, we really have to look at people and what
they took with them and where they went.
HOPPER:
And who worked with whom.
TROPP:
Right.
HOPPER:
Now you, for instance, you'll find in those early days at UNIVAC that a number of
people came and took those early courses and then went out from there. The whole
prudential group ( ?). The John Hancock group, those groups that came into
UNIVAC for training and then went out to spread it in other places.
I go around the Country and then someone will come to me and say, you don't remember
me, but I got my, you taught me about computers in 1950. You don't remember me, but I
heard your lecture in 1950. This is the way it spreads you see, from those early things.
And it spread through people. Not the colleges until much later.
And there was no publication… (voice fades out). I can walk in to any manufacturer
today and point to the people I trained originally. That's why I said when I (got the
( ?) award ?) that the most important thing I could do was all the young people I
trained, not the development of the compiler.
TROPP:
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
44
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
When you say you took the ideas from Harvard that led to the compiler, what were the
kinds of pressures that you were feeling, while you were programming at Harvard that
ultimately led you?
HOPPER:
Well, for instance, it was the development of all the pieces of coding that were written in
relative code. And it was pretty quickly obvious that you could process those and piece
them together into one program and that's all the first compiler did.
Plus my own laziness and natural refusal to do anything over again that I had already
done. And the fact that when I got down to UNIVAC they were building BINAC then
and I had to learn how to program it. So I learned how, I still thought I was a
mathematician.
I learned how to add, subtract, multiply and divide. I was real good at it, the only trouble
was, at the end of the month my check book didn't balance.
And I had to realize, my brother, after three days of hard work, said, the rumor was not to
subtract in octal. (LAUGHTER).
I had to realize that I couldn't work eight hours a day in octal and then live the rest of the
time in the normal decimal world. And my answer was not that I learned better octal, as
the damn computer could learn decimal. I would instruct it in my own language. They
could do the dirty work. The transformation because it was built to do that. So I …(voice
fades out).
And in my own case, very great deal of practicality and common sense. I had no great
development abilities or creativeness or anything. I had a very large amount of common
sense and a very great deal of (Von Neumann's views and a lot of them?) that I had
already done.
TROPP:
There are some of the characteristics of Mark I that I'm not sure why they were there. For
example, you mentioned the twenty three digits and I wondered if this isn't because some
of the early impetus for Aiken back in the '30s wasn't from people in astronomy who saw
needs to do computation.
Because otherwise I don't know why he had so many digits in the …(voice fades out).
HOPPER:
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
45
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
Any engineer had to cope with vessel functions. The only way to compute them was by
recursion. And you've got to have a lot of digits for any of the recursive formulas and he
eventually did the vessel function.
TROPP:
So that you're speaking of the set of tables that he, (in reliance?)?
HOPPER:
Recursion. The phenomenon of … The only …
TROPP:
And that's why the machine was built so that you could get thirty six digits if you needed
them?
HOPPER:
Well no. That was added later. That wasn't in the original machine. The double precision
was added later. Much of the computation was done by recursion. These were the only
formulas we had and if you were going to use any kind of a reverse formula, you've got
to start off with a heck of a lot of digits.
TROPP:
I guess that's right because that, Aiken says that in his '37 paper, that this has got to be the
…
HOPPER:
And if you look at the early matrix work and the early partial differential equation work,
you had to start with twenty three digits or you weren't going to come up with anything.
We had to have it. Or we wouldn't have had the answer, we would have blown it.
If you remember Babbage had forty some digits in his differential (calculator?).
TROPP:
I guess the reason I asked the question was because in the, early in the late '30s when the
machine was not yet designed or off to IBM for building, two of the people at Harvard,
that were pushing hardest were members of the Astronomy Department.
But I don't know of any incident, or problems that were later solved.
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
46
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
HOPPER:
No. I think it was largely the matter of the recursive formulas. And not much
investigative work had been yet done on how much accuracy lost in computation and
there was a tendency to put in plenty, so you would be safe.
TROPP:
So your round off error ( ?).
HOPPER:
Yes, in truncation error. They didn't know what was going to happen on truncation errors.
They had no idea. And a function like the sign can roll off the machine awful easy. So
that, I think that was there for safety.
I know, on the first early problems, when we were computing how much accuracy we
were having, it was an appalling rough computation. We didn't know much about it yet.
And I was unusual because most mathematicians didn't know anything about round-off
errors and truncation errors and the reason I did was because I had taken a course on
chemistry, that was where I learned it.
TROPP:
That's interesting.
HOPPER:
And that was where I had learned about round-off errors and computational errors, not in
mathematics.
TROPP:
How did you work in ( ?) and statistics?
HOPPER:
Not in graduate school or anything, I took a course in chemistry… Mathematicians didn't
worry about digits anyway, they used symbols. Probability wasn't that far along. Statistics
wasn't that far along. Extremely elementary at that time.
TROPP:
I wish we had the time to get off into the later work with UNIVAC and BINAC. We
probably ought to save that for another time. (LAUGHTER).
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
47
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
HOPPER:
I'll come again. Incidentally, feel perfectly free to call me up if you've got a question at
some time. Just give me a ring.
TROPP:
Let me turn this off.
HOPPER:
Because I'll be in my office the whole of July.
LUEBBERT:
You were in charge of Mark I?
HOPPER:
Later.
LUEBBERT:
When Mark II and Mark III were being developed?
HOPPER:
After everybody else began to go play with Mark II then they left me in charge of Mark I.
Which was a victory on my side because when I walked in there he had not wanted a
woman officer and I had said he was going to want a woman officer.
You see, in 1946 for instance, the Navy turned me down for regular navy. I was two
years too old. It took me twenty years to win that one. (LAUGHTER).
I kind of make a habit of winning. It's taken five years with this triple standard.
TROPP:
Well you know, it says something then for Navy discipline in regard to Howard Aiken,
he accepted the fact.
HOPPER:
He had to. He was in the Navy too. I think probably the greatest disappointment of
Howard Aiken that he ever faced was that he didn't make Captain and that was due to one
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
Computer Oral History Collection, 1969-1973, 1977
48
Grace Murray Hopper Interview, July 5, 1972, Archives Center, National Museum of American History
For additional information, contact the Archives Center at 202.633.3270 or archivescenter@si.edu
very bad fitness report which had been written by an officer ( ?) and he
never made Captain and he should have.
But he didn't, and I suppose quite accurately from that officer's point of view, an
insubordinate and the fitness report was in his file and …(voice fades out). He should
have made Captain.
Dostları ilə paylaş: |