688
SERGIU HART
come apart. So that aspect of my thesis—which is the easily formulated part—did
survive.
A little later, but independently, Dick Crowell also proved that alternating knots
do not come apart, using a totally different method, not related to asphericity.
H: Okay, now that we are all tied up in knots, let’s untangle them and go on.
You did your PhD at MIT in algebraic topology, and then what?
A: Then for my postdoc, I joined an operations research group at Princeton.
This was a rather sharp turn because algebraic topology is just about the purest of
pure mathematics and operations research is very applied. It was a small group of
about ten people at the Forrestal Research Center, which is attached to Princeton
University.
H: In those days operations research and game theory were quite connected. I
guess that’s how you—
A: —became interested in game theory, exactly. There was a problem about
defending a city from a squadron of aircraft most of which are decoys—do not
carry any weapons—but a small percentage do carry nuclear weapons. The project
was sponsored by Bell Labs, who were developing a defense missile.
At MIT I had met John Nash, who came there in ’53 after doing his doctorate at
Princeton. I was a senior graduate student and he was a Moore instructor, which
was a prestigious instructorship for young mathematicians. So he was a little
older than me, scientifically and also chronologically. We got to know each other
fairly well and I heard from him about game theory. One of the problems that we
kicked around was that of dueling—silent duels, noisy duels, and so on. So when
I came to Princeton, although I didn’t know much about game theory at all, I had
heard about it; and when we were given this problem by Bell Labs, I was able to
say, this sounds a little bit like what Nash was telling us; let’s examine it from
that point of view. So I started studying game theory; the rest is history, as they
say.
H: You started reading game theory at that point?
A: I just did the minimum necessary of reading in order to be able to attack the
problem.
H: Who were the game theorists at Princeton at the time? Did you have any
contact with them?
A: I had quite a bit of contact with the Princeton mathematics department.
Mainly at that time I was interested in contact with the knot theorists, who included
John Milnor and of course R. H. Fox, who was the high priest of knot theory. But
there was also contact with the game theorists, who included Milnor—who was
both a knot theorist and a game theorist—Phil Wolfe, and Harold Kuhn. Shapley
was already at RAND; I did not connect with him until later.
In ’56 I came to the Hebrew University. Then, in ’60–’61, I was on sabbatical at
Princeton, with Oskar Morgenstern’s outfit, the Econometric Research Program.
This was associated with the economics department, but I also spent quite a bit of
time in Fine Hall, in the mathematics department.
INTERVIEW WITH ROBERT AUMANN
689
Let me tell you an interesting anecdote. When I felt it was time to go on
sabbatical, I started looking for a job, and made various applications. One was to
Princeton—to Morgenstern. One was to IBM Yorktown Heights, which was also
quite a prestigious group. I think Ralph Gomory was already the director of the
math department there. Anyway, I got offers from both. The offer from IBM was
for $14,000 per year. $14,000 doesn’t sound like much, but in 1960 it was a nice
bit of money; the equivalent today is about $100,000, which is a nice salary for a
young guy just starting out. Morgenstern offered $7,000, exactly half. The offer
from Morgenstern came to my office and the offer from IBM came home; my wife
Esther didn’t open it. I naturally told her about it and she said, “I know why they
sent it home. They wanted me to open it.”
I decided to go to Morgenstern. Esther asked me, “Are you sure you are not
doing this just for ipcha mistabra?,” which is this Talmudic expression for doing
just the opposite of what is expected. I said, “Well, maybe, but I do think it’s better
to go to Princeton.” Of course I don’t regret it for a moment. It is at Princeton that
I first saw the Milnor–Shapley paper, which led to the “Markets with a Continuum
of Traders” [16], and really played a major role in my career; and I have no regrets
over the career.
H: Or you could have been a main contributor to computer science.
A: Maybe, one can’t tell. No regrets. It was great, and meeting Morgenstern
and working with him was a tremendous experience, a tremendous privilege.
H: Did you meet von Neumann?
A: I met him, but in a sense, he didn’t meet me. We were introduced at a game
theory conference in 1955, two years before he died. I said, “Hello, Professor
von Neumann,” and he was very cordial, but I don’t think he remembered me
afterwards unless he was even more extraordinary than everybody says. I was a
young person and he was a great star.
But Morgenstern I got to know very, very well. He was extraordinary. You
know, sometimes people make disparaging remarks about Morgenstern, in par-
ticular about his contributions to game theory. One of these disparaging jokes is
that Morgenstern’s greatest contribution to game theory is von Neumann. So let
me say, maybe that’s true—but that is a tremendous contribution. Morgenstern’s
ability to identify people, the potential in people, was enormous and magnif-
icent, was wonderful. He identified the economic significance in the work of
people like von Neumann and Abraham Wald, and succeeded in getting them
actively involved. He identified the potential in many others; just in the year I
was in his outfit, Clive Granger, Sidney Afriat, and Reinhard Selten were also
there.
Morgenstern had his own ideas and his own opinions and his own important re-
search in game theory, part of which was the von Neumann–Morgenstern solution
to cooperative games. And, he understood the importance of the minimax theorem
to economics. One of his greatnesses was that even though he could disagree with
people on a scientific issue, he didn’t let that interfere with promoting them and
bringing them into the circle.