An Interview with Edward Teller

Edward Teller Edward Teller


GAM = George Michael



GAM: It was June 30, 1995 and I was going to interview Edward Teller. Quite unintentionally, this interview turned out rather differently from all the others. It happened this way: About two weeks before the date of the interview, on the recommendation of Edward's secretary, I sent a memo to him outlining what the interview was about and some of the things we wanted to cover. So, the morning of the interview, I was ushered into his office and he smiled and said, "Why are you coming here?" I said, "I'm here to interview you." He said, "What about?" I said, "I sent you a memo describing that I was trying to record some of the history of the early days of the computer department, and since you were very active then, I was thinking that you might have something to say. He looked at me and said, "I don't remember." Then he added, "I didn't really have much to do with the establishment of the computer department. Sid Fernbach did most of that, but I will tell you a story about Johnny Von Neumann. I happened to be in Washington, D.C. for some sort of a meeting and Johnny called me to his office saying he wanted to talk about the UNIVAC LARC, which we were then trying to procure. When I got to his office, he said, "I think you're asking for too much memory. You don't need that much memory for a computer." We were trying to get twenty or thirty thousand words. I was too smart to argue with him, and I just left and came back to the lab. We talked about it at the Lab and decided that we had to talk to Johnny at the Lab. So, I called him up and asked him to come out so that we could talk about this problem. When Johnny arrived, we showed him one of the design codes that was being prepared for running on the LARC. We asked him, "Can you make some suggestions about how we might fit this into ten thousand words, or less, since you are saying we don't need any more words than that? We know that if everybody was as smart as you are, we would already have figured that out, but since we don't have those kinds of people here, we'd like to ask you." Well, the long and the short of it is that Johnny withdrew his objections and we were free to get the amount of memory we thought we needed. Ultimately, the LARC was delivered late, but it had thirty thousand words of memory on it.

GAM: The LARC memory was available in multiples of ten thousand words. Each ten thousand-word piece was housed in a cabinet that was about 8 feet tall and about 16 feet long and about 3 feet wide. So having thirty thousand of such words meant that we had three long cabinets that filled the LARC room. The majority of the cabinet was filled with driver circuitry and the memories worked well, but they were huge.

Anyway, that ended the interview, with Edward telling me that the strategy they followed was essentially to get Von Neumann out here and show him a big code and ask him how to shrink it to less than ten thousand words. Obviously, he saw that that was not possible, so he withdrew his objection, and we were able to outfit the LARC with the amount of memory we thought it needed.