Video Excerpt

GEORGE
At the end of the war, everything began to dissolve. People went their own ways, which meant getting out of the military and going back to civilian life. I too was looking for a job after I finished my Ph.D. I was looking for an academic job.

When the war ended, I went out to Berkeley to finish up my PhD. That was one semester, and then I came back to the Pentagon. Berkeley made me an offer, but I didn't like it because it was too small. Or, to be more exact, my wife did not like it. It was a grand salary of fourteen hundred dollars a year. She did not see how we could live on that with our child David.

I returned to the Pentagon and began to look around for another job. The date was June 1946. I was in the market for a better paying job at a University. I received a number of offers. My colleagues at the Pentagon wanted me to stay on and were busy looking for something that I might do to keep me in, to keep me interested. The idea they came up with was to mechanize the planning process. Before that, in the Pentagon, it was all done by hand.

IRV
So you decided to stay on at the Defense Department and to work on mechanizing the planning process. One of your first papers was entitled "Programming in a Linear Structure". What were the mathematical structures that you had in mind at that time?

GEORGE
The mathematical structures were what we called later linear programs. Our immediate goal was to speed up what was being done by hand.

IRV
In June 1947 you came up with the linear programming model and then you proposed in August 1947 the simplex method of solution, correct?

GEORGE
Yes.

IRV
How long did it take before the military was able to realize some of the benefits of the models and the algorithms that you were proposing?

GEORGE
About eight years. You must remember in 1947, computers were just beginning to be built and it was not until 1955 that they began to be reliable. We implemented around 1950 a mathematical model which we called the "Triangular Model" which was solved using punch card machinery.

IRV
Can you tell us about the first applications of linear programming that were solved by hand?

GEORGE
There were two.

There was something called the Berlin Airlift. At the time, we were having trouble with the Russians, and they had blockaded Berlin. So the American and Allied forces, other than Russia, were like prisoners in the Middle of Berlin. The only way that the supplies were coming into Berlin at all was via airplanes. The United States and Britain had control of an airfield in the Berlin area, and we were flying in supplies. There was one program that we invented called the "Berlin Airlift", which set up the logistics of it, and we solved that problem.

We also solved Stigler's "Minimum Cost Diet Problem". The hand calculations were done by a team under the supervision of Jack Laderman at the U.S. National Bureau of Standards.

Previous

Next