An Interview with Jim and Mary Lou Moore

JM = Jim Moore
MLM = Mary Lou Moore
GAM = George Michael

GAM: Today is April 23, 2001 and I've planned for years to interview the pair of you. I thought it was a great idea but it's never happened until now. We're just going to remember some war stories; it isn't documented history or anything like that and I'll ask you, Jim, to start.

JM: I graduated from UC Berkeley in February 1953, and I went to work at the Lab that March. They had just started the lab about six months before that, I think, in September of 1952. What attracted me was that I heard the Lab had bought a UNIVAC, the first commercially available, large�I guess you could then call it large�digital computer. During my senior year at Berkeley, I had the good fortune to take a class given by Turbin Misling. He was a computer scientist in the early 50s whom UC Berkeley stole from MIT. He taught a basic computer class. It rang a bell with me. I could tell it was going to go far, so when I found out the Lab was getting a UNIVAC, I applied. At that time, they were looking for people like crazy. They invited me with open arms, I was pleasantly surprised. I was initially assigned to the maintenance crew on the UNIVAC.

The UNIVAC was a 5000-tube machine, its mean time to failure less than twenty- four hours, as you probably remember. We had a four-shift operation, Sid Fernbach said, "Don't let that machine sit idle, have it doing work all the time." So it ran seven days a week, twenty-four hours a day. I stayed with the UNIVAC, until about 1955. And then the Lab went out to bid for a supercomputer. Supposedly pushing the art. There were two bidders, IBM and Remington Rand's UNIVAC. Remington Rand got the contract and they proposed the Livermore Advanced Research Computer (LARC).

GAM: I should say that we also called it the Leith Atmospheric Research Calculator because it was one of the first things that ran on LARC.

JM: Right. Well, anyway, I wasn't really involved with writing the specs. In fact, my boss at the time was Lou Nofrey. Remember Lou?

GAM: Yes. Let me tell you a sad aside. I was on my way to Oroville to interview Lou when I got word that he had died. I was very saddened by this. In those early years, he helped me in so many ways; he was a great engineer and friend. He was a great guy.

JM: Yes, his death was very sad. I think he was one of the prime movers for that contract as well as a lot of others. But, anyway, Remington Rand got the contract and they actually stole away quite a few of our people.

GAM: Larry Harrison is the only one I know.

MLM: Bob Crew, did you know him?

GAM: Yes, I knew him. I didn't know he'd gone there, I thought he stayed with Lou.

JM: Oh yes, he did but, of course, you knew Nofrey went to Remington Rand too?

GAM: I didn't know that.

MLM: He didn't actually go to Philadelphia with the LARC. He went to CDC in St Paul.

GAM: I don't think CDC existed then.

JM: Yes, that was a splinter from Remington Rand. But I think he was in Philadelphia for a while because I remember seeing him there. However, I think he was involved with planning; he wasn't involved with the LARC. Bob Crew was. In fact he was one of the mechanical designers for the drum system.

At one point, Remington Rand offered to train us to maintain the LARC after we took possession, offered us the opportunity to come back there and help the budget. This was good for them, because it was sort of free labor, although we didn't know too much. Anyway, that's where I met my wife. Mary Lou was one of two designers for the CPU of the LARC and I got to know her pretty well. I got to know her so well I decided I'd like to marry her and she agreed. Isn't that amazing?

GAM: That's amazing, yes.

MLM: The truth is, they told him to bring back one of those girls from Philadelphia.

GAM: Why don't you jump in here and say where you came from?

MLM: I went to a small Catholic Girls School in Philadelphia that never heard of a computer.

GAM: What was the school?

MLM: Chestnut Hill Collage, just a little coconut. But when it came time to look for a job, everyone was out interviewing and I already had a couple offers to teach school; I just went to UNIVAC to interview for fun. I just wrote them a letter and said, "I'm graduating with a degree in Math; what can I do?" I was interviewed at IBM and at Remington Rand. IBM was unspectacular, but Remington Rand offered me a job at $330 a month.

GAM: That's more than the Lab started us at. That's good.

MLM: And then, that was our job, to design the LARC. We knew nothing; we didn't even know about binary arithmetic, knew no physics, nothing.

GAM: Well, the LARC was decimal, but that's OK.

MLM: Right, right.

GAM: Well that's great, I actually spent fourteen months at Haverford and Bryn Mawr and Swarthmore Colleges.

MLM: That's right, I did hear that about you. When you teaching CCD at St. Michael's.

GAM: Yes, I was for a while.

MLM: And that was before you went east. Well anyway, in those days, maybe only Berkeley had a class in computers but they sure didn't on the East Coast.

GAM: That's been a common thread through all the interviews. There was no Computer Science.

MLM: There was no training. When people interviewed me for jobs at Remington Rand, I asked, and they said, "Well, nobody has any experience, nobody has any education. We just find out if they can think, if they can play chess, if they can play bridge, maybe if they majored in Math."

GAM: So you really had no computer-related background when you started work?

MLM: That's right! It's scary when you think about it. They sent two girls who didn't know anything about how to design this thing.

GAM: Well, it's clear that you did a very good job.

JM: I think that was the intent, they wanted someone who wasn't jaded and simply going to invent the past. That's what I heard. Anyway, there wasn't enough experience for anyone to be jaded. We were surrounded by UNIVAC people, who thought like UNIVAC people. And that is, I think, what they wanted to change.

JM: I'd like to digress here because something you said reminded me that I was interviewed for the job at the Lab by Jim Norton. It turns out he was the head contract administrator for the LARC, so we knew each other and I essentially felt like I was working for him during contract negotiations. We had a lot of meetings; it seemed like they went on for many, many months. We'd go back to Philadelphia and we'd have a meeting and see how progress was going and then a month later Remington Rand people would come out and have a meeting with us and tell us what they were doing. Anyway, at one point, Norton was getting concerned about us taking care of the computer after we took it over and he told me, "Make sure you bring one of the designers out. And so I did. Of course, I had plans to do that before he said that.

GAM: That's great. He didn't tell you to marry her, but you certainly do have good taste.

JM: I like creative engineering. However, that became a convenient way to do it. And I've got follow-on to that, too. We took delivery in June of 1960 and several people from Remington Rand stayed with it for a few months to make sure we were comfortable with it. I don't think we were ever comfortable with it. But problems would come up and we would solve them one way or another. I remember you mentioned anecdotes. A problem developed, and I don't remember the exact details, but it was in the CPU and the machine was absolutely down, it would not run and one of my favorite techs was Jack Noonan. He and I worked on that. I don't know how many hours we worked trying to find the problem, Mary Lou called me and said, "When are you coming home to dinner?" because it was starting to get late. And I said, "Oh, we've got a problem out here and I'm not sure when we'll be done and I briefly described to her what we were seeing and she said, "Oh, it's probably ASTZ." That's a signal. I said, "Oh, no, it can't be that." I didn't quite know what it was. But, anyway, Jack and I had done quite a few hours of diagnosing and, sure enough, the next place we looked was this ASTZ signal, and sure enough, that was the problem. I don't remember the exact details, ASTZ has something to do with generating all zeros in the CPU. I almost didn't want it to be that.

GAM: I understand.

JM: I fought it for a while, but we finally had to admit that was the problem.

GAM: What you should do is take her out for a Champagne dinner. I think that's great, remote debugging.

JM: Right, there was very little description of the problem and she hit it on the head. Like I say, we'd been working on it for hours. I thought that was very funny. I've never forgotten that.

GAM: Well, I don't think you should. It's an excellent story. Now, as I heard things, you were in charge of all the people who went back.

JM: Right. The maintenance crew.

MLM: Ten guys.

GAM: Did that contribute to your gray hair?

JM: A little bit, yes. A little bit, definitely, there were some problems.

GAM: When I started at the Lab in April of 1953, the UNIVAC had been delivered essentially a week earlier, but it didn't come up and run until about six months later.

JM: I didn't think it was that long. I know the maintenance, or the installers were there for several months after I got there.

GAM: Well, I think it's six months because everybody was writing a thing that Tom Wilder started called the "Three Word Input". It was a program to help the UNIVAC keep track of it's own inventory. And we didn't get to test it for a long time, six months about.

JM: Yes, could be, could be.

GAM: I have a picture somewhere of the UNIVAC Birthday Party. I guess it was in 1959 or so. I don't know that the machine ran after 1959.

JM: I don't think it did. But that was a fairly long life, nine years.

GAM: I think, maybe, the thing that's more impressive was that out of, essentially starting with a blank slate, they put together a whole UNIVAC system. Before, there were just isolated pieces of equipment. After, there was a computer system. True, some parts didn't function too smoothly, but nonetheless, it was a system. I remember sitting there watching Ray Teleski clearing errors at the console, doing the SCICR thing hour after hour. I don't remember now what SCICR stood for, but it involved lots of manual operations at the console. Our program ran forty hours on that machine. At least it could run that long before people lost interest in it, so we had a lot of SCICRs. I got to be very good friends with Dick Carpen also as he replaced worn out 25L6 tubes. He was a great person. So anyway you brought the LARC back and you got it running but you were also present at the death of the UNIVAC.

JM: Well, no, I wasn't associated with it anymore. I'd gone to work in EE, for a short time, and supported a couple of other programs that weren't really part of Computation. But then the LARC contract came up and I got back into Computation with that.

GAM: The interesting thing about the LARC is that it was supposed to be the first computer that used transistors. But its impact was very strongly attenuated because, since it was late, transistors had been first introduced in other machines that we bought before the LARC arrived. We were using the CDC1604 and the IBM 7094 by then and they were using transistors. IBM had a lot of trouble with their STRETCH; they had the same trouble that the LARC had earlier.

JM: Right, the STRETCH was delivered a few months after the LARC I think.

GAM: Yes. And you know from the A-Division point of view, STRETCH paid for itself during the test series in 1962. So did the LARC.

One of the things I'd like to get into, if you guys remember, there's a story that Nevin Sherman was unable to find agreement in the 22nd decimal digit of his astronomical orbit calculation.

MLM: Ah, the logic error in the LARC.

JM: The double-precision multiply. That was a pretty subtle error. We worked on it for quite a while. It was very strange. I remember that you would print out, on a line printer, and about the third or fourth digit over in significance there would be two fives, 55, right next to each other. And Nevin had checked it out and it should have been 56, but it was printing 55. And so one of the things that hung us for a while was, we figured, that there are these little hammers that activate the characters and there are like eighty of them across. And we were figuring instead of hitting the six, the original five, which was the correct one, was pulling the other five along with it and we figured there was no problem in the CPU at all, just a problem with a sticky printer. That diverted us for a while. But we soon ruled that out, but I don't remember the details of the fix at all.

MLM: It was because we had parallel digits and we were using two or four words with an exponent of one. And the carry net was as saturated as it could be. And so, when we added these two together, there was a sign-digit which got blanked out and the carry didn't go from the end of this word to the beginning of the next word. It should have been an eight instead of a nine or a nine instead of an eight. It was a design error. Like I said, I've forgotten the details.

GAM: That's ok; I have the details captured in both the Noonan and Sherman interviews.

JM: Sherman was the one who discovered i,t and Noonan was the one who finally set it right. Mary Lou was raising children at the time so we didn't get her involved.

MLM: The day I brought my first baby home from the hospital they called from Remington Rand, to ask if I would please edit some journals. They called Jim also. Who was the guy from Berkeley, in electronics? He said why didn't I come to work for him. At the time, husband and wife couldn't work in the same group, and he said, "Well, you could work at Berkeley." And I said, "Commute to Berkeley?" And he said, "Well, move to Danville and then each of you can commute." And I thought, well, that sounds like a solution.

JM: Oh, wasn't that Norton? Jim Norton?

MLM: He was the one that was the big shot in charge of the LARC.

JM: Yes, he was a contract administrator. He was first in Berkeley. As soon as the LARC contract was done, I never saw him again. He went on to something else.

MLM: Of course, it all just fizzled, because Remington Rand didn't have the courage to make more machines. They should have had one following in the production process and they didn't. They were afraid.

GAM: Well that's probably true. In fairness, I think they had spent so much money building the first two that they didn't want to spend more money on it.

MLM: But, the trouble was, they used it developing the first two. If people only spend their money on developing and never put something into production, they'll never retrieve their money.

GAM: I understand that, but they wanted orders and there were none forthcoming. The only place that even came close, as far as I know, to ordering a LARC, was UCLA. And that fell through because they decided to go for the binary STRETCH. In retrospect, all of us were blindsided by the business of decimal versus binary. Since we cut our teeth on the UNIVAC, we thought that was real, that's what God wanted.

JM: Yes that's right. Ten fingers.

MLM: Well, they had a lot of odd ideas. Like they put about twenty percent more equipment in to handle the check digit. And all that did was to introduce twenty percent more chance of error. So there were a lot of unchallenged assumptions.

JM: You mentioned UCLA, I was interested in that because I went down to a course given by a professor on computers and he was a decimal-machine advocate. I bet he was involved in UCLA's interest in the LARC.

GAM: Well, you know, the "Natural Physicists" think that a tenth ought to be a terminating decimal. So, that should be true in a computer as well. Like God favors the decimal number base. Ah, well, so much to learn about that sort of thing.

MLM: Well certainly. We were just about ready to go into physically building the LARC, having designed it and everything, and Press Eckert sent a question to us. He wanted to know how much it would cost to make the LARC into a binary machine. Of course, it was such a humongous cost that the idea was abandoned.

GAM: A complete LARC redesign?

JM: Yes, you couldn't just snap to and tomorrow the job would be done, and so it just never got further. But he did think of it.

GAM: Was he more influential then say, Herman Lukoff?

MLM: Well Herman worked for Press Eckert. Press was an odd, odd person. Lukoff was a steady, steady; Mr. Sure. Nothing flaky about Herman. But there was lots flaky about Press Eckert. You know Press would bring his eighteen-year old girl friend of the moment through the plant at night and show all his things. And, in those days, there were a lot of experiments set up; they had things like disks set up with strings and pulleys. And he'd go through the labs and show off for her and take things apart and then people would come to work the next morning and their experiments were in pieces all over the place. Somebody went to his office once and he was sitting cross-legged on his desk breaking pencils. Taking the secretary's pencils while she was taking dictation, breaking it in half and throwing it across the room and she had to go throw it in the trash. So he was strange. And, while I was there, he was promoted to a position of non- importance. Remington Rand never fired anybody so they promoted you sideways. Instead of being in charge of the whole Philadelphia Plant, all of a sudden he was on the same level on the chart, but there was nobody under him. And that happened to Press Eckert.

GAM: That was when Lukoff took over?

MLM: Lukoff, no. Chon Chou took over.

GAM: I heard that Lukoff tried to move the entire project to Saint Paul.

MLM: Well that could be. He didn't have that big a power..

GAM: Well he wanted to do it just the same.

MLM: Herman was an Engineer.

GAM: He wrote a book titled From Dits to Bits. He was a radio enthusiast; an, amateur radio ham. Then he went into the military and became a radio expert and then he came out and went into computing.

MLM: He used to call his wife on the way home in the court and tell her he was coming and the Philadelphia police picked him up for calling on his radiophone. I think now of the cell phone people. Poor Herman got a ticket for doing that. But he was too nice, too ordinary of a person.

JM: Well, he was a nuts and bolts down to Earth guy, Herman was. And Press Eckert was kind of an idea, blue-sky guy.

MLM: They were eccentric and sparky.

GAM: Well he was not as blue-skyish as Mauchly was, but...

MLM: Well, Mauchly was a different brand of visionary. Eckert pushed Mauchly out. Another one that pushed Mauchly out was Grace Hopper. Grace Hopper and Mauchly were doing the programming and Eckert was doing the engineering and Grace, pound for pound was much more fierce than Mauchly and she wound up being in charge. She really pushed Mauchly around.

She used to come around, she was a reserve Navy gal, you know, and we worked in an old Battery factory back there. Dirt was the name of the game and she used to come out with her white gloves and walk along. It galled my boss, he also was Navy reserve and it galled him that she outranked him in the Navy reserve.

GAM: I officed with a wonderful person by the name of Dana Warren for the first few years at the Lab and he was married to a woman whose name originally had been Grace Adams. And Grace Adams and Grace Hopper roomed together in College. So whenever Grace Hopper came out here she'd go visit the Warrens. We'd all go over there and we'd each take a book and read a Shakespeare or a Bernard Shaw play and she had this sort of high pitched voice and she was sort of strange and very opinionated. We told her about the project that Merritt Elmore had to build a compiler for output on the UNIVAC. You write something like FORTRAN format statements and it'll print your stuff out for you. But somebody classified the output because it had words instead of letters so it got to be more readable, I suppose is the way you'd put it. Grace pushed that a lot though. She went over to COBOL, but I didn't like that at all.

MLM: I worked with people, well one I'm sure was a genius, Bill Schmitt, who was responsible for the logic and Art Gehring was in charge of design. But some of it was because they had the freedom to be free, so to speak.

GAM: There is that, but the managers had not got to the point where they were organized enough to manage anything OK? So, we had a lot of freedom everywhere, up and down the line; a very productive time for everybody.

MLM: My daughter is working at SUN now and she said they were talking about how they are going through this listing of most valuable to least valuable people in the each department and merging them. And in her department there's a particularly eccentric guy who has gotten a couple of bad things written on his file. They need him but now that puts him on the bottom of the pile and makes him the first to go. And she said her boss is very alarmed because you can't throw out all the people like that because that's where a most good ideas come from.

GAM: At least her boss has the wit to recognize that. It doesn't happen often. Well, go on with your narration now, we keep interrupting you.

JM: Let's see, I'm trying to remember when I got offered the LARC. It was about 1964, I think.

GAM: Apropos of that, the LARC had certain alterations on it. I know I participated on one of them.

MLM: Oh, they put IBM tapes on my LARC.

GAM: Yes, we had IBM tapes, and I think I participated in the business of correcting the EPR's, the Electronic Page Recorders. They were a product of Stromberg- Carlson and those guys didn't have enough respect for what a power supply could do when you use it. So everything would sag and the character lines would be wavy and you couldn't expose the film the right way and all that stuff so we threw that stuff out and we put in new cameras and CRTs and a new character generator, and it got to be a fairly useful output device.

MLM: I remember when I saw those IBM tapes, and I thought, "Oh man, this is the end."

JM: That actually extended the life of the LARC I think.

GAM: The LARC introduced many new ideas. One of the most seminal, in a sense, was the register files. That idea reappeared in the PDP-6 designs of Ben Gurley at DEC. He came and saw the LARC, he said that's a good idea and that's what he did with the PDP-6. And I can tell you another anecdote that is cute. I interviewed Teller and he said he didn't remember anything except one thing. When he was in Washington DC, Von Neumann grabbed him and said, "Your wasting money on the LARC. You'd only need 10,000 words at the most, 4,000 would be enough. Edward said, "I was too smart to argue with him." But he came home and organized himself a little bit. He invited Von Neumann to come out for a design review or something like that. Basically the message came through in the review, that well, if you're Von Neumann you can get along with 4,000 words, but ordinary people need much more. So Von Neumann withdrew his objections and we got 30,000 words of memory, eventually, and that satisfied Von Neumann. He never objected. He was influential on the General Advisory Commission of the AEC then and he could have stopped us getting enough memory. I thought we should get the 100,000 words, but we didn't do that.

JM: Yes, well each 10,000-word cabinet took a lot of space, George.

GAM: It did indeed.

JM: That was one of the things, unfortunately, that Remington Rand felt they couldn't do was to solve transistor drivers for the memories. They did use tubes as the drivers for the memories. Driving cores takes an awful lot of power and transistors just weren't ready to hack it yet, so that in fact was the most least reliable part of the LARC, the memory drivers. They used transmitting tubes, great big ones consuming 20-50 watts.

GAM: If there'd been 100,000 words on that machine I bet it would have lasted at least another ten years.

JM: Maybe so. Well, like I said, I went on to do other stuff. I got involved with the PDP-6 and the PDP-10s. I added a Data Cell to the PDP-6, which became a medium-capacity kind of storage. It was really a mass storage device. It held 3.2 x 10^9 words. At the time that was quite large. The right unit, you remember. That's a 36-bit word.

GAM: It actually held more than that, but they were in 32-bit units because of the nature of binary addressing and things like that. I gave the museum over at Moffett stuff from the Data Cell. And I gave them four-foot platters from the Bryant files from the CDC 6600, and film cells from the IBM 1360 Photo Digital Store, stuff like that. They're all over there. We have a thing called The Memory Lane and it's devoted to the development of memory storage. The Data Cell was first in this lane.

It's pretty interesting. They have another lane over there that's devoted to the computers of Seymour Cray. He, on his own, decided he needed something to control his robotic stuff so he bought some Macintoshes and then took them apart and understood how they worked and those were the things that controlled the wires and the component inserters and stuff like that. Because you couldn't get a person to do it, but you could get a person to steer it. That was very nice. So he had some T-shirts, I don't know where these came from, but they said "My other computer is" and if you were an Apple employee, you'd put a CRAY on it and if you were from CRAY you'd put a Macintosh on it.

Well, when I visited them in Colorado Springs, they had dozens of Macintoshes. Each one working on an assembly making something. What a genius he was in packaging, wow.

JM: That was his secret with the CRAYs�the ability to get the heat out. That ECL stuff, was a real big advantage. You remember MASS right? The MASS Storage System?

GAM: Yes indeed, the CDC 38500 Cartridge Store.

JM: Sam Coleman was one of the primary programmers, remember that?

GAM: Yes, it came after the Photo Store. The CDC cartridge actually had a better form factor than the IBM 3850 cartridge, but it was just too complicated.

JM: It was complex, yes. That's when I got a chance to work with Sam. I really liked working with him.

GAM: Yes, he is quite calm and relaxed, and a very good programmer.

JM: Right.

MLM: But then you really stopped doing engineering and just did programming?

JM: No, I was still doing design stuff. Remember when we went out to bid on the high-speed printers? IBM got the bid with the 3800. I was involved in putting the vector generator scan converter on that. And there was a fair amount of software that was devoted to producing graphics faster than if you did it at dot matrix, much more files. And so we did the same thing when we got the IBM laser printers. It was 32 inches a second and we used 8 1/2 inches, so divide 8 1/2 into 32. 3.76 pages per second; almost 4 pages/sec.

JM: One of the big problems, it seemed to me, with those high-speed printers was just handling the paper.

GAM: The radiation printer handled paper beautifully. It was built on a printing press frame.

JM: But you had to use a forklift just to stick a new roll of paper on it because it weighed over 200 pounds.

GAM: Yes that's true.

JM: Well, like I say, I got involved with the PDP-6 then, which was going to be sort of a head for the OCTOPUS System and there were lots of devices that had to be put on it, the Data Cell being one of them. We needed more registers; I added some registers to it that were really coupled to the CPU very tightly.

GAM: So, you are on the PDP-6 and the PDP-10 and you're putting all these things on, it. That was essentially giving birth to the OCTOPUS.

JM: I'm trying to remember what came after that. MASS came after that. That was a fairly long-term project. We had to design channels to connect to the PPUs of the 6600's and 7600's. John Randolf and I worked together. He worked the PPU end on the 6600's and the 7600's and I worked the MASS end, and we designed a channel where all the channel hardware in both our devices was identical and then we put it in interfaces. In my case, an interface to MASS and in his case an interface to PPU.

GAM: What was the bandwidth on that channel? Do you remember?

JM: 42 megabits. We had to go faster than the PPU disks and they were about 40.

GAM: 42 million bits a second?

JM: Right. We used high quality cable. But we needed that for a good signal from the MASS.

GAM: Well, but the thing that was looking at it from the MASS side was this TI-980 or something?

JM: Right. Sam Coleman wrote the system code.

GAM: Right. I thought he did a beautiful job.

JM: Oh, he did.

GAM: But I didn't believe the machine could go that fast.

JM: Its problem was it didn't have a stack. That's what Sam kept complaining about. It had an ability to save all the registers in one command and he liked that. But a stack he could have used. So he tended to favor PDP-11's.

GAM: But, after the adventuring of the MASS what did you do?

JM: Then I got involved with Systems Concepts. Remember Systems Concepts?

GAM: Yes. I thought they were wonderful.

JM: They were. They were very much DEC-oriented and they offered great equipment. I guess we bought some tape drives from them and I got involved in handling that contract with them.

GAM: They sold us memory. And they sold us the entire data gathering system at Nevada. And it was built on a PDP-10 like frame. That's what I wanted here, but Sid got the LDS-1 from Evans and Sutherland instead. Also a very good machine, but it was never used.

JM: Right. That's too bad, but Sid had his way.

GAM: Sid was absolutely good at dealing with people. And that's a fairly uniform thread that goes through the interviews. Wyman thought he was great and he gave examples why. Sid knew how to treat people well. He wasn't a super manager on a project. But he knew how to handle people very well.

JM: I enjoyed the dealings I had with him very much.

GAM: After he died, Borchers asked me to pull together a memorial seminar in his honor. So I did and they liked it quite a bit. I had the help of Ralph Carlson and we had maybe a hundred people show up for it. Hans Mark was there. And there were a lot of others. Even Edward Teller came. Everybody told stories about Sid and about future things. I didn't have the wit to tape everything. It would be nice to compare with what they thought and what we've got right now. I think we've gone off into the left field right now. More and more people are starting to write analyses of the kinds of problems that we had to solve at the Lab. They appear in weather and in crashing cars, in nuclear design and stuff like that, and they don't fit at all well on the kind of machines that the Lab is buying. You may have to completely redesign to fit on that new IBM thing. What I hear is that nobody cares anymore about efficiency. So, these problems are showing in the order of about one percent efficiency or less.

One percent! But who cares, it runs. The most popular machine here is the ALPHA Cluster that we started getting from DEC and handed up, dealing with COMPAC who bought DEC and they got all of the research and the research labs at Hudson and things like that. And now they are in a head to head competition with Dell to be the biggest supplier of personal computers.

Anything that lowers the price will get the American's interest. Irrespective if it's any good or not. But I also hear that many of the weather researchers in this country who are trying to do some serious computing must go to Japan or to Germany to run on a machine fast enough for the new models. There are some attempts in the US to regain the forefront of High Speed Computing but right now Japan has it.

In a way, Japanese ascendancy can be dated to after 1980. I think there were two reasons, they stayed at the job of exploiting vector processing, and they didn't cave in to money problems. So they built the machines and if they didn't perform too well, they revised them and rebuilt them.

Here everybody was trying to undercut Seymour. He was the only one who was doing good designs. Now we have Burton Smith with his TERA. His company bought CRAY from Silicon Graphics. And so he has the CRAY management structure and the maintenance structure that he can get at. Plus he's got his own; he calls his design Multiple Threaded Architecture.

You get two hundred and fifty six processors. They're fairly robust and they go fast, I mean relative to the teraflop range. But you know they talk about all the tasks his stuff needing to get to the next three orders of magnitude higher, petaflops.

Well, I'm sorry, I keep interrupting you. So what happened next? You were doing some procurements with systems concepts.

JM: Right, systems concepts. We needed high speed, high performance tape drives for the PDP-10s.

Actually I worked on the Sigma-7 for a little bit too. The Sigma-7 had limited memory and they needed more. I can't remember how much I added, but this was in the days when CORE was still king. And the Lab had gone out to bid and got a whole bunch of 16K memory modules from AMPEX and I added a couple of those and had to interface it to the Sigma-7. We also had a bunch of high performance graphics consoles, but they had to be in the room with the Sigma-7.

GAM: Well, there was one down at the other end on the Sigma-2.

JM: Well, there was a task but it never really came to fruition, but I was involved in designing a system that would allow local monitors down in 131 on the Sigma- 2. I don't know if they wanted to get rid of the Sigma-2 or what, but we got to the point where we had the channel going. Then the software effort was pulled out from under us and the Sigma-7 went out.

GAM: You know the Sigma-7 represented a lot of development. There were some very good guys on that project.

JM: Like Kelly Booth.

GAM: Kelly, and Charlie Wetherall, and Gary Anderson was there. I don't remember any of the others. They built a very good time-sharing system called GORDO, but I haven't been able to find anyone to talk about it.

MLM: PDP-70's, when was that?

JM: That was late, Mid 1970s.

MLM: When was Carver Hill around?

JM: He had left by then. Carver came about the same time Bob Wyman did, mid 1960s. And he left, I think, in the early 1970s.

GAM: Wyman says that Carver found a notice that the school where he had gotten his Bachelor Degree was offering an MD degree for anybody who had a Ph.D. in anything else. So Carver went back and became an MD, and that's what he is now.

MLM: Yes, but he wanted to be a doctor. He had been working in an Emergency Room for fun. H.e had been volunteering there. And he had a new wife and a new family. Three small children plus the two from his previous marriage. He started over a couple of times. He left the Lab and got his Ph.D., and then he started his own company and then sold the company. It looked like he was going to be a student forever. So when he went to medicine, I pictured him out at Stanford sometime. But he's a family Doctor in North Carolina.

He came out here when he was looking for residency interviews and I said to him, "Wow, Carv, isn't it hard to go back and study?" We had a daughter that was doing the same thing. He said, "Nah, Mary Lou, there really wasn't much to it." And I think that life experiences, plus he's a really smart kid, just made it all mesh, it must have been a pain for all these young kids who were television scarred, to study and have this guy just lay back and say, "Well now, I don't know it, but it's easy."

GAM: So, anyway, you became more of an Executive, right?

JM: No, not really. I wanted to do design work and the last task I had was the Vector Generator and Scan Converter for the IBM 3800, the laser printers. And that was a fun task. I designed the hardware for it and we had a fellow in our group who was going to do the software because it required some micro programming in the vector generation and converter. But he got pulled off and they said to me, "You're going to do the micro programming." So I had to do that too. And I was a little leery of that but it turned out to be a lot of fun. I really enjoyed it and it took me to my retirement.

And I think it's strange now too. I think how on the cusp everything was then and I had not the slightest inkling of anything. The boss always asked, "Why don't you want to be on a committee?" But I said, "it just isn't there." We went down to San Louis Obispo when one of the kids was in Computer Science down there and they were having a show, The History of Computers, and well, there's my computer, there's the LARC. It's funny to be part of history.

MLM: When I took classes at Chabot out here, I took some FORTRAN classes. The kids in the classes were all young kids and they said, "What was it like, what was FORTRAN doing?" and I said, "There wasn't any FORTRAN when I programmed." And they couldn't imagine anyone that old. You know, "NO FORTRAN?" In the class were about thirty-five young computer science majors, a math teacher from Livermore High, a young kid from high school in Dublin and myself. And the first night was three hours and he whizzed through number bases; octal, decimal and binary. And when I came home I said to Jim, "If I hadn't had all that background I wouldn't know what that man had said." And he went through the second night and then he gave a quiz. And he says to me, "You know, nobody understood a word I said, they practically panicked, they didn't even know enough to ask a question." And they had just shot down six hours of class time. So, he regrouped and started all over again.

GAM: Did you run into Pat Crowley out there? That's more recent. Pat was teaching BASIC out there.

MLM: Yes, they didn't have BASIC when I was out there. Bob Dickinson got the first little PC. I can't even think who made them. It came with no directions and so he had a couple of these eager beaver young technicians, who'd hover around and try this and try that to get this thing working. And other than that, they had a PDP something in Hayward. And, so, he ran the classes through that. And the same with Pete. In Pete's class, we did homework on the PDP.

GAM: Is Nicholas out there yet?

MLM: Oh no, I don't think so, jeez that was a long time ago. It might have cured him of teaching because some things you can take in big gulps, three hours one night a week, but not something as intricate as programming. You have to get a little bit and use a little bit and get a little bit more. It's like math; you can't take an afternoon of math and then not look at it until the next afternoon, because if you're behind, man, you're shot. And that's what he discovered; he really had a lousy situation.

GAM: I'm really interested in your appraisal of what happened to your career. You finished an under-graduate major in Mathematics and went directly into one of the most difficult parts of Logic Design?

MLM: Well, I don't think that's true. I think the essence of the design of the computers, then anyway, was your ability to break things down into "yes" or "no". You had to start with you are or you aren't. The light was on or the light was off. And it was, in essence, making things simpler not more complex. And then you chose to take those simple things and join them up. So, it really wasn't that hard and it was a lot of fun. It was like being paid to work a puzzle, and we got to do it all. In the morning, ten people in the design department at Remington Rand would meet to discuss the status of our work, so everyone was clued in. We talked to the STRETCH designers, for instance, at a convention and they had, maybe, fifty-five people. And that's harder, because you can't be so sure that everyone is up to speed.

And we had people like Larry Harrison working with us. You know he was a nice guy. And they were just nice people. And Crew was a big friend of Larry's. When he came back they saw a lot of each other. It was just a fun time. And it was kind of sad to see it have to be profitable. It wasn't profitable, so it didn't last.

GAM: To some extent, no early computer was profitable with the possible exception of the 704. I was told that the IBM 704 was largely a commercialization of the SAGE computer. You know the ideas were all worked out there. Pioneering business.

MLM: Art Gehring, who was in charge of the design department had worked on BINAC and other stuff and, of course, Preston had worked on the ENIAC, but otherwise we had nothing to do with SAGE.

GAM: OK, I believe we're at some sort of a finish point. I want to thank both of you for taking the time to discuss your careers. Believe me when I tell you that your stories are wonderful and surely will please many who read them. Thanks a bunch.