An Interview with John Ranelletti
JR = John Ranelletti
GAM = George Michael
GAM: Today is July 10, 2001. I'm going to talk to John Ranelletti today. John, why don't you start by telling us from where you came, how you got to the Lawrence Livermore Laboratory and your early experiences there.
JR: All right, George, I'll be happy to do that. A little bit of history on our family side. In 1961, I married my wife Barbara and we moved from Southern California to Northern California, principally because I had been admitted to the University of California, Berkeley Graduate School for Physics. I started taking classes in September of 1961, and by the fall of 1962, I had decided that I'd had enough of Graduate Physics for the time being. We had a house Castro Valley, and my wife had started teaching high school, so I decided to drop out of the UCB program and go to work. I made some applications, and I received two responses in October of 1962. I got a lot of interest from the Lawrence Radiation Lab (LRL) in Livermore and also a response from Lockheed, Sunnyvale. Since the Lab response came first, I called up and made arrangements for an interview. In the fall of 1962, I went out to the Lab, where I was hosted by several of the Lab people at the time: Hans Bruijnes, Richard Von Holdt, and Tad Kishi.
GAM: Dick was a great Mathematician.
JR: Yes, and Joe Brady, they all spent some time with me. They ushered me around the lab, giving me a tour, and I had lunch there. I spent the whole day at LRL, and I was so impressed that when I got home I said to Barbara that if they'd hire me, this is the place where I want to work. The Lab had rooms full of computers and all this fancy stuff. At that time, my only computer experience had been during one summer, while an undergraduate at the University of Southern California (USC), at the Southern California Gas Company, where I learned to program the IBM 650. It was a 2K drum machine. It was a big thing in those days, and they were using it for gas meter reading predictions. While learning to program there, we did our programming in SOAP. In fact, the operating system was called "SOAP-2." The system optimized programs on the disk, so that as the disk rotated around, and you'd get the next instruction quickly. This saved compute time, because the instructions were optimally placed around the storage disk.
I had limited knowledge at the time about computing, so I went into the interview at the Lab as a wide-eyed young kid, and I said, "This is really a cool place." I kind of embarrassed myself during my interview at lunch with Von Holdt, because to try to make an impression as a young person, with my limited wisdom of the day, I said, "Do you program here using floating point or do you use only integers?" That immediately showed my total ignorance of the subject. But nonetheless, they sent me a hiring letter, and since I hadn't heard from Sunnyvale, I decided that I would grab the opportunity to work at the Lab. On January 28, 1963, I joined the Lab as a "Mathematical Programmer 1" in the Computation Department, with a starting salary of $625 per month. I was hired by Sid Fernbach, as everyone was in those days, and initially, I was assigned to the "Cooler," which was a two story military barracks style building located where the Physics Building currently stands. It was Building 415 as I recall. At that time, the building housed people waiting for their security clearances. I was ushered in there, and they gave me a desk. Initially, I was the only one from computations in "cooler area," because I started in January, and there had apparently been a lot of people who had received their Q-clearances the previous fall. So, I was there by myself. About two weeks later, Frank McMahon showed up and joined me in the Cooler, then Ira Morrison, and then they added additional new hires.
Ed Schoonover was my first supervisor; he came over with an armload of FORTRAN manuals, and he said, "Read these," and then he disappeared. So I read the FORTRAN manuals, and he'd show up and answer my questions. That's how I got started. Well, I spent six months in the Cooler. That was not an unreasonable time in those days; you got cleared in about six months. I was cleared in June, 1963. My first "real" assignment was given to me in the Cooler, because I was learning FORTRAN: To work on a code that was part of the PLOWSHARE program at the time, the PLUTO code. It was a FORTRAN code that simulated part of the PLOWSHARE process, and it was enormous in size ... it was huge! My job was to convert it from the IBM 7094 to run on the LARC. I knew nothing about the 7094, I knew nothing about the LARC, and I didn't know much about FORTRAN. But, I thought, "what the heck, I'm here to learn," and I jumped into it. I really enjoyed it. When I got my security clearance in June, I was moved inside the Lab, and I was assigned to Building 116, the old two-story Computation office building. I had an upstairs office, which I shared with Roger Moore. I was basically working as what they informally called a "Junior Physicist." I continued programming (applications programming), on the project for PLOWSHARE. As I worked away on this program, I would submit my code changes for keypunching. Once, I received the new punched cards, I'd take them to the LARC and submit a job for a LARC run. I remember that I'd often have some "crazy compiler error," and I would get, and this is not an exaggeration, a "hand truck load" of memory dump pages from this machine. It was five feet tall or something, it was incredible; the system just printed out everything in the computer memory. So that was a good incentive not to make an error! At the same time, I was using the radiation printer, which we had in those days. It was very fast, thirty thousand lines a minute. Seven pages a second, fast for it's day. Fast for today, actually. And it was a special "one of a kind" printer, and it was basically a gray surfaced, black-backed paper that I can remember always came off on the white shirts that I wore in those days. I remember distinctly getting a call one day from computer operations, Mona Millings. You remember Mona? She used to run the high-speed printer in those days. She called me up and she gave me holy hell, because I had programmed a "1 in column 1" (which meant new page). The result was that I sent this printer into a printing frenzy producing a page with only a the number 1 in the upper left corner, for pages and pages and pages. As a young programmer, I was learning the ropes by making mistakes, but I still enjoyed it. I really liked the environment; I met a lot of good people. Around that time frame, I met you, George, for the first time, when you dragged me over to see the PDP-1 and EYEBALL experiment that you were doing. I thought that was really cool; there was a large CRT, plus all the cool graphic things that you were working on.
I continued working on this large program and getting to know the LARC, and I was learning to use the FORTRAN compiler by translating what the IBM 7094 would do. I enjoyed my work. One day, there was a knock on my office door, and Hans Bruijnes walked in with Ed Schoonover. Hans, in his unique style, said, "Have you had enough time fooling around with this stuff? You want to do some real work?" What he offered me was the chance to join a group that Sam Mendicino had started, "FORTRANing FORTRAN." They were getting ready for the CDC 6600, which was coming in, and we'd just had the delivery of two CDC 3600s. Hans wanted to know if I would be willing to work in this new group, because they needed help to get this FORTRAN FORTRAN version off the ground. It sounded interesting to me, so I agreed, and I immediately transferred over to Sam's project. I worked on a package called "IOH," which was the I/O FORTRAN package. It was the software interface between the user's FORTRAN code and the buffer I/O software that drove the tapes and the disks. I started by inheriting a set of subroutines from a guy who had just left the Lab. He wrote the IOH software in FORTRAN, and after I picked it up, I continued to develop it in FORTRAN as well. It ran horribly slow, and my task was to get it to run fast. We sped the software up by trimming things down; I worked with Sam, who was doing the compiler. Along about that time, in the fall, Rich Zwakenberg joined the group, when he got out of the Cooler.
Sam built up a group of about half a dozen people, including Bob Hughes and Jeannie Martin, to work on the FORTRAN project. Also there was a competing effort, in those days there was more than one system group. There was another group that was doing an alternate approach of putting a system on the CDC 6600. The GOB system then was being developed by Tad Kishi, Bob Abbott, and Norm Hardy. Norm worked on the communication between the PPU (peripheral processor) and the central processor (CPU). Thus, there was a little bit of competition at the time; we were working on the FORTRAN implemented system, and they were working on the GOB Operating system, which was written in assembly code. The effort turned out to be a real blessing for me, because I really enjoyed working in the systems area. It was a lot of fun and I met a lot of interesting people, Norm Hardy, Bob Abbott, Tad Kishi, and all the people in Sam's group.
When the CDC 6600 arrived, we had the opportunity to put this FORTRAN FORTRAN compiler on the machine. Initially it ran slow, but we worked on it to improve its efficiency. I got the IOH software running pretty well. At the time, I received a lot of assistance from David Storch and George Sutherland. Also, I worked closely with Doug Kent, who did the buffer I/O software, Lee Tennant, and George Powles. The IOH software would feed calls to Doug's routines, and he would do the actual I/O data transfers. It turned out that we actually got to the point where we ran at acceptable speeds. Around 1964, the first CDC 6600 arrived. I had the opportunity to meet Seymour Cray for the first time when I went back to test some of our compiler software on the machine. I remember walking into the CDC Chippewa Falls lab one evening (they gave us time at night). There was guy sitting in the corner punching cards on the keypunch in a Lumberjack shirt, and it turned out it was Seymour. He whirled around and introduced himself, and he showed us how to get started. We spent the evening debugging our stuff. The next day, we met with Seymour and his staff, and then he took us out to lunch. I remember to this day, I was sitting across the table from Seymour Cray, and he was asking about what we were doing. I explained to him that we were "FORTRANing FORTRAN." He asked, "How many people are in the group?" I said, "Oh, there's six," and he looked at me and he said, "That sounds hopeless." By his way of thinking, that was way too many people. In Chippewa, he had one guy doing the compiler, one guy doing I/O, one guy doing the operating system, plus himself. That was his whole software staff! He was somewhat amused that we had six people working on something that he thought one person could handle, that was his style. So we did the FORTRANing FORTRAN, eventually as time went on the CDC 6600 machines arrived, and then the CDC 7600s came. FORTRANing the codes and the operating system became the distinction for us at Livermore. We benefited from the fact that we developed everything in a high-level language, and we received credit for that accomplishment both inside and outside the Lab. That was an exciting time, because we did it all. We started with the raw hardware, and we built everything, all the way up through the applications and eventually the graphic interfaces. The initial networking consisted of PPU communication with the CPU on the CDC 6600. I think we did it well. That's the thing that, for me, distinguishes those days from today. In those days, we were implementers and creators. Today, I think, Computation acts more like integrators; they integrate system software obtained from the outside. So I really look back fondly to the 1960s.
CDC, in it's wisdom or lack thereof, it's hard to say, (I maintain it was a step we had to go through), decided they were going to design the STAR computer in the late 1960s. I had, by then, along with the other members of our group, developed quite a bit of experience developing compilers, and we were going to put the FORTRAN compiler on the STAR. The STAR-B machine was the first model. Also, Rich Zwakenberg and I had an idea that we could use PL/1 as a programming language at the Lab. We wanted to see if we could put PL/1 on the machines at Livermore. So, he and I together, actually wrote a PL/1 compiler, and it was the first PL/1 compiler ever to run on a non-IBM machine. It was great experience, because in those days you had an idea, they would let you go do it, and we did it. We did this PL/1 implementation in addition to the work we were doing in preparation for the arrival of the STAR machine.
Along about 1969, Bob Abbott approached me; he had quit the Lab, and he had gone to work for a start-up company in Berkeley called "Berkeley Computer Corporation" (BCC). Professor Mel Pirtle and a couple of other UC Berkeley Professors in the Engineering Department started the company. They wanted to put together a hardware version of a timesharing system. They designed the BCC 500. Its claim to fame was that it could hardware-allocate portions of the computer hardware to individual users. The plan was to offer timesharing in a commercial environment. Bob Abbott was hired by them to build the applications area, and he was looking for people. He asked me to join him at BBC. I was young, and a twenty-percent raise and the change seemed appealing. I had worked at the Lab, at that time, for seven years, and I'd just turned thirty. We had two young kids (Scott and Eric), and we were building a new house in Hayward, California. Barbara and I discussed the pros and cons of the job offer, and I said, "You know I love the Lab, but I'm not sure if I want to work there my entire life. I want to see if I can do something outside and this is an opportunity." In retrospect, it was the "dot-com opportunity of 1970." So, I told Bob I would accept his offer, and I went in to talk to Sid Fernbach, the Computation Department Head. Sid said, "Oh, you're making a big mistake. Those guys are big trouble. Trying to hire you indicates they are in even bigger trouble. Don't go." I told him that I was committed to go at that point. Sid said, "No, don't go. What do you want to do here? Maybe we can make some changes." I responded, "No, I have to go." So I left the lab. Well, Sid was right, they were in financial trouble, and they lasted about three months after I joined them, but it was an interesting three months. Shortly after I arrived at BCC, because of my background, they shifted me from the applications to the systems work. I was helping to develop a QED Editor to run on their machine, and I worked with people like Butler Lampson and Peter Deutsch and several other Berkeley people who were very bright and stimulating individuals.
I enjoyed working at BCC a lot. Judy Ford had also joined Bob Abbott at that point in the group, and we worked at night. We had time on the Shell Oil machines, time-sharing off the Shell computers at night. I worked fairly close with Peter Deutsch at the time, because he was sort of my mentor to get me started on the QED Editor at this time. It was an exciting time. About three months after I arrived, they called us into a meeting and said they didn't have enough money to keep going and that the few people that were the founders were going to stay on and try to do something. Eventually they got the BCC computer at least to the point where they could ship it to the University of Hawaii, as a research machine.
I was a young guy at the time, and we had this brand new house on the hill in Hayward (where I'm still living today) and two young sons, and I said, "I really can't stay." So I called the Lab up, and I talked to Sam Mendicino. Sam said, "Well, why don't you come back?" And I said, "Thanks, but I've only been out three months, and I haven't really seen what I can do outside." At the time, I happened to hear that there was an opening at Control Data Corporation. CDC had a small research group in Palo Alto: The Control Data Research Group. An Ex-Lab fellow worked there; he wrote the compiler for the LARC, while in Livermore.
GAM: Lee Krider?
JR: Yes. I heard that he was looking for somebody. I had known Lee at Livermore in my earlier days. So I called him up, and he said, "Indeed, I'm looking for somebody." I went over to meet him in Palo Alto, and I learned he needed someone to join him to write a compiler for a hardware project that Control Data was investigating. It sounded really exciting to me, so I joined him at CDC. I started commuting from Hayward to Palo Alto, and the CDC facility was just down the street from Stanford. It was a small group, and it was a unique opportunity because there was just Lee and me working on the FORTRAN compiler. The majority of the group were in Minneapolis. Control Data had asked Semour Cray to build a 6000 PPU Cluster without a CPU. The question at the time was: could you make ten PPUs work together without a CPU to create a system and run an application, basically the forerunner of multiprocessing.
I was really excited about this project, so Lee Krider and I created the FORTRAN compiler. There was another guy who did the operating system, and another guy who did the I/O. It was kind of like Semour Cray's model, you know, three or four people working on a big project. We got the multiprocessing system up and running. It ran so well that if you didn't have heavily sub-scripted variables, with more than two dimensions, the cluster of PPUs ran faster than the 6000 CPU.
GAM: Really?
JR: Yes. Since Lee was the project manager, I eventually ended up having control of the compiler. I had it to a point where you started compiling and, if you put your main routine first, we would start executing your program before we compiled your sub-routines. So the source code compilation would go through and start executing, and then if the code called a routine that wasn't compiled yet, the program would wait. Once the called routine was compiled, the program execution would resume. It was just something to prove that we could do such things with this new system. We were really proud of our new system, but Control Data decided there was no future in it. They canned it. Then the corporate management said, "You have to find something else to do." An invitation, of course, was given to go back to Minneapolis, but I didn't want to do that. So I looked around locally. In Sunnyvale, Control Data had a larger group, and they were developing applications based on this thing called the "Zodiac Operating System." The Zodiac Operating System was a new concept in operating systems for Control Data. They had recently received a contract with the Bank of Switzerland to put four hundred terminals in Switzerland. I told them I was looking for work, so they invited me to come to a meeting in Sunnyvale. When I walked into this auditorium full of several hundred people, I looked around, and I said to the person next to me, "Why are all these people here?" He responded, "This is the software group to work on operating system." I said to him, "You're kidding me." And he said, "No, that's true, this is as many as we could get today, and we have another crew coming in tomorrow for the same briefing." I said, "This is impossible." Well as it turned out, I heard later that this project did not do well. The effort got too big and too cumbersome. I didn't go with them. I happened to know a fellow whom I worked with in Palo Alto, who also went to Sunnyvale. His name was Mac MacDougal, and he wanted to develop an "event based simulation language." At the time, Control Data had a contract with NASA to provide a database for the data that was collected by Skylab. Skylab would send the data down, and it was retrieved by two IBM machines on the front end. The data was then fanned out to five Control Data machines. The idea was to find out if you could simulate this whole process. I joined Mac, and in three months, Mac and I produced a brand new simulation language. I programmed it in FORTRAN, because of my experience, and Mac designed the simulation constructs. At the end of that three months, since I had decided prior to working with Mac (and he knew it) that I was going to go back to Livermore, I returned to the Lab. After I looked at the big Zodiac Operating System project, I said, "I can't do this." At the time, I had been out of the Lab two years, and I decided to go back. The Lab said they would hire me back, but I told them I didn't want to come back until I got my security clearance reinstated. So I agreed to work with Mac on his simulation language, for the Skylab project until the clearance arrived. We developed the language called "ASPOL."
We got ASPOL working in three months, and then my clearance came through, so I left Control Data. Looking back, I remember that three months as one of the most productive times in my career; working with Mac was a great experience. So, I went back to the Lab in March of 1972. Well, when I came back, Sam didn't have an opening in his group, but fortunately as it turns out, I went to work in John Fletcher's group. John Fletcher had a group of people working on various system projects. One of the things that they needed help with was on the PDP-10 computer. They wanted to implement the DEC operating system on top of the lab operating system we had, because there were some applications users that wanted to use the DEC operating system in Livermore. I rewrote an interface, with the help from other group members, especially Garret Boer. I had to learn the PDP-10 machine language to do this project. The interface functioned so that it would look like a DEC operating system, even though it was running Livermore's operating system. We got that going pretty well. In Fletcher's group, there was another fellow named Clarence Badger; he and I eventually were assigned to do the systems software for the first CHORS (Computer Hardware Output Recording System) project. It was a high-speed printing station for all the supercomputers at the lab. That was a great time too, because I had an opportunity to do work in assembly language for mini-computers. We developed the whole software system, Clarence and I did. The system sat downstairs and received the print data from the machines upstairs, the CDC 7600s and the CDC 6600s, even some of the IBM 7094s that were still left. From the upstairs computer's perspective, we were a magnetic tape drive. We couldn't change the hardware upstairs, so we had these boundary conditions to work with; we had to pretend we were something that they knew about upstairs. The CDC 7600 PPUs would write a special channel, as if it were a tape, and downstairs we would collect the data with our operating system. We would spool it out onto disks, and then retrieve the data and re-send it to the high-speed printers, which were purchased from Honeywell. We had to get the data off the disks and into the printed page of memory each time the page went through, so that we would get printed images. It looked to the Honeywell printer as if the data came from a tape, and from the CDC 7600 or the CDC 6600 it looked like the data was written to tape. So, all the machines thought they were reading and writing tape, but our specially developed software and hardware was in the middle.
Clarence and I did a software implementation, and in those days it was exciting, because all the hardware interfaces had to be built inside the Lab. There was a piece of hardware called "The Teacup" which was a device that sat on the end of the upstairs machines that pretended that it was tape. There was a large buffering hardware device for all the vector generation of the characters and graphics. All the stuff we had to do get the printed material into the page buffer for the printers had to be done in house electronically. There was about a half-dozen very talented engineers and technicians assigned to our group. The design engineers: Don Wentz, John Randolf, Jock Vogelander, plus several others were there. We all worked together. Clarence, as the lead software person, I, plus the engineers put the CHORS-1project together, and it worked very well! I enjoyed working on that development team very much.
If we can step back a for a moment, to 1972 when I returned to the Lab, I was working in John Fletcher's group, but I had historical ties to the Language Group. During my absence from the Lab, Mary Zosel had joined the Language Group. Mary came from the University of Washington, and she had some interesting ideas about programming languages. She designed a language called ALA, which was a PASCAL-like language, that we were going to try at Livermore. So, in addition to working in John Fletcher's group, I worked with Rich Zwakenberg and Mary to implement the ALA language, which was an experimental effort. In the process of doing that, I was invited to teach a class for the Applied Science Department of the University of California, Davis.
The Department of Applied Science (DAS) is a branch campus for UCD. There was a course entitled: Introduction to Programming Languages, and I was asked to teach it. I taught it for a half a year (one semester), and I really enjoyed teaching. While I was teaching this class, the Computation Department became part of the Chemistry directorate under Gus Dorough, and there was a re-organization of the Computation Department in 1972. The department was split into four divisions. Sid Fernbach was still Department Head, but four new division leaders were appointed. One of the divisions was the Computer Systems Division. The other three divisions were Operations and two for applications.
After approximately 27 years of Sid's leadership, this was a major change in the structure of the Computation Department. The reason I mention this is because, when I was working on the CHORS Project in 1976, Sid Fernbach, for whatever reasons, stepped down or was encouraged to step down, from his department head job, and they appointed a new department head, Bob Lee. Shortly after Bob Lee came into the department, Dave Pehrson, who was leading the Computer Science Division, left and went back to Engineering. Bob Lee decided to split the Computer Science Division into two parts, the Networks Systems Division (NSD) and the Users Systems Division (USD). They posted the two new division leader positions, and some of the people I worked with approached me, and they said, "Why don't you apply for one of these division leader jobs?" As of this time, I had had no leadership experience at all. In fact, I had avoided it; I had spent my entire career at that point staying away from anything that looked like management. However, I got to thinking about it, and I said, "I haven't tried that sort of thing, but it was a unique opportunity," so I submitted my name for consideration as the Division Leader of the Users Systems Division. At the time, it was a group of approximately 80 people, and they developed and maintained the system and mathematical software that wasn't the operating system. The Network Systems Division people did the operating system. Lo and behold, Bob Lee appointed me to be the USD Leader, and he appointed Sam Mendicino to be the NSD Leader. Suddenly I was thrust into the new environment of "management." I took it fairly slow initially, because I wasn't sure what I had to do, but we needed to keep the division moving.
There were five groups in the division: Language, Graphics, Publications, Computer Research, and the Library Group. We had to pull all this together and develop the software that was needed to support the users on the current systems in Livermore, which was LTSS at the time. Thus, my first experience as a manager was in the role of division leader, and I stayed in that position for four years. I learned a lot about working with people during that time. When Bob Lee accepted the position as Computation Department Head, he told Gus Dorough, who was the Associate Director at the time, that he was only going to do the job for four years. Bob Lee was true to his word, and four years (and three seconds) after his appointment, he stepped down. Now they needed to search for a new Computation Department Head, and since I had been doing the division leader job for a while, I was encouraged to apply for the Department Head position by several of the people in USD. I decided that I would apply. There was a search committee appointed by Gus Dorough, and George Michael was a member of this committee. As I recall, there were several applicants, all of which I thought were very good, at least the ones I knew about.
To my surprise, Gus Dorough appointed me the new Computation Department Head. This appointment completely changed my career from a technical person to a complete manager. Actually, the division leader job partly accomplished that transition as well, but as the division leader, I tried to keep my hand in the technical work. I assigned myself to work the User Help Desk, just to keep track of the issues, but I couldn't do that as Department Head. One result from becoming the Computation Department Head was the job put a halt to my work towards a Ph.D. in Computer Science at the UC Davis Department of Applied Science (DAS) in Livermore. In 1975, after teaching the programming language course for UCD, I enrolled as a graduate student at DAS, working toward the M.S. and Ph.D. degrees. I had completed all the course work, and I had started to investigate a thesis topic when the department head opportunity came along. With the appointment as Department Head, I pretty much stopped all work on the Ph.D. I held the position of Department Head from 1980 to 1989. My predecessors; Sid Fernbach had the position for twenty-seven years, and Bob Lee had the job for four years. Bob Lee, in retrospect, was a transition between Sid and myself. In the 1980s, it was a completely different world than it was in the 1960s and 1970s. We had lots of pressure from the user community, and the government itself, to get more into the role of "systems integrator" rather than the role of "systems developer."
We started the 1980s with a big buy of computers that was orchestrated through the Gus Dorough's AD office, when we acquired five Control Data machines in one "big buy," one for the Magnetic Fusion Energy Computer Center (MFECC), two for us in the Livermore Computing Center, and two for Los Alamos. It was an exciting time, and as I have mentioned, it was very different than in the 1960s and 1970s. I viewed my role as department head primarily as managing the department itself. The AD's office, at that time, had pretty much taken over the procurement areas. Sid Fernbach was on the AD's staff for a while, and eventually he retired. Bill Masson and Gus Dorough handled the outside procurement issues, and I had the inside department issues. That worked fine for me. I enjoyed the role of managing the department. It had its challenges, since by then it was a department of some seven hundred people. So the challenge of the time was to try to balance the requirements of the systems development against the needs of applications development and support. When I became Department Head, there was a feeling amongst a good portion of the members of the department (the fastest growing part of the department: the applications area) that they were "second class citizens" relative to the systems people in such things as assignments, raises, department support, etc. We worked hard to try to address those feelings, and whether we were totally successful or not, I'm not sure. The point is that there was a time in the department where a lot of emphasis was put on system development in the 1950s, 1960s and 1970s. Then as the tenor of the times changed, applications became even more relevant to the use of the machines at Livermore, because we were using commercial software, and we had to continue to develop the applications. So, that was the time in the 1980s when the applications people, I think, rightly deserved an equal share of the pie, whatever that means. So I, hopefully, started the process, so that we could evolve to equity within the department.
It's not clear, in my mind, that we were totally successful, because there were long histories there. We did what we could do. I was Department Head for nine years, and about seven years into that tenure, I received a letter from the University of California, Davis, saying, I was still on the books as a "Ph.D. graduate student," but I was not making any progress. This was true. In effect, the letter said: "you either have to shape up or ship out." In other words, I had to decide whether to finish my degree or not. I went home, with the letter, and I said to Barbara, "You know, I've got this department head job that's occupying a hundred and ten percent of my time. I really don't need this degree." And she looked at me and said, "You're stupid, you better finish it. You've put all this effort into the degree, go ahead and finish it. If you don't finish it, you're going to feel bad about it later." So I took her advice and, it turns out, it was excellent advice. I spent two years finishing my degree while I was Department Head. I worked eight hours a day as Department Head and, since I had a second office I could assign myself, I spent the evenings working my research. I did that night after night, and I eventually finished the degree. Garry Rodrique was my committee chairman, and Jim McGraw was my thesis advisor and technical mentor for my research. My dissertation is entitled: "Graph Transformation Algorithms for Array Memory Optimization in Applicative Languages". To this day, I'm thankful for Barbara's support and advice, because completing the degree was an important accomplishment for me. In 1987, when I finished my degree, one memorable experience for me, personally, was that the whole Computation Department had a party. They invited me into the auditorium under the pretext that I was giving an award ceremony for some people who had earned service awards. Every time someone got a service year pin, we tried, rightly so, to make a big deal about it.
GAM: Yes, you had a reputation of being a very clever MC.
JR: I enjoyed giving attention to people who worked a long time at the lab. When a person puts in twenty or more years, they deserve to get their moment in the spotlight for sure. So, anyway, my secretary, Virginia Dubose scheduled me to go the auditorium, because we couldn't get Building 113. I thought that was strange, because I thought the auditorium was way too big for our service awards ceremonies. We usually had about sixty to eighty people attend. When I walked in, the place was full of Computation people, and they had a surprise reception for me for getting my degree. My family was there also, Barbara, our children: Scott, Eric, Darin, and Marla, plus my parents. That was an especially nice thing that happened to me.
Anyway, in 1987 or around that time, the Computation Department actually got its own Computation Directorship, and Bob Borchers was appointed the first AD for computing at Livermore. At that time, that was a new thing. It was a large recognition that computers had come a long way in status at the Laboratory.
So we got our own AD-ship, and Bob Borchers was the AD and I enjoyed working with him. He came in with the idea of learning what we did, and then he supported us. He did that quite well. So, for the two years or so that he was the AD, he was looking at ways to best organize the computation effort at Livermore, plus he had responsibility for the Magnetic Fusion Energy Computer System (MFECC) as well as the Computation Department. He decided he would reorganize the Computation Department, and he created two organizational entities that reported to him, which comprised the applications and computing science areas, and the systems area. As a result, he appointed me to be his Principal Deputy AD, along with Dieter Fuss who was his Deputy AD for operations of the Computations Directorate. So Dieter and I joined Bob in the AD's office, and that arrangement lasted between 1989 and 1992. Along about 1992, Bob Borchers stepped down for the AD position, because there was a lot of politics going on about how to handle computing at the Laboratory. He went on to run the University Relations Office. Bill Lokke was appointed, by the Laboratory Director, as the Associate Director for Computations. Bill came into the job of AD with the idea of building his own staff, and consequently he basically told me that I had the opportunity to look for something else to do. In retrospect I can't fault that, because when you come in as an AD you might want to appoint your own Principal Deputy Associate Director. I had not done technical work since 1976, and now I was faced with the prospect of doing technical work once again. I happened to remember that Chuck Cole had mentioned to me, about six months previously, that he had received funding from the Office of Classification in Washington D.C. to do a pilot project on declassifying documents using software. Because that was related to compiler techniques, and somewhat related to my language experience and my thesis in the sense of potential algorithms and analysis, I called up Chuck and asked him, "Do you still have that funding. Is the position still open?" He said, "Yes," so I took the job. I left the management position, and I went back to full time technical work, which was a welcome change for me. It's a little like riding a bicycle, you do technical work, and then you don't do it, but when you get on this old bicycle again, it wobbles a little bit, and then you kind of start down the road, and pretty soon you kind of pick up speed, and you learn the new stuff, and it all comes back to you.
So, I worked four years as the Principal Investigator for TAP (Text Analysis Project). A team of three of us at Livermore (Susan Taylor, Mary Schrot, and myself), along with an outside consultant (Huy Joy), developed a software package that could analyze English text to identify classified concepts. Of course, it was a classified project. Our Text Analysis Project was a prototype for work that's actually continuing in Washington D.C. today. They are using different approaches, but we provided the proof of principle that, in fact, you can use software to detect, computationally and with reasonable accuracy, classified concepts. After we proved the principle that the text could be analyzed, the funding eventually ran out for the project. At that point, I decided I would apply for a job in the Advanced Scientific Computing Initiative (ASCII) Program within the Livermore Computing area. I joined the group to work on support tools and parallel processing and memory tools for the ASCII System. I worked with Brian Carnes (Group Leader), and I was reunited with Rich Zwakenberg, with whom I had started at the Laboratory years and years before, and Mary Zosel, plus several of the other people whom I had known in the past. It was a little like "going home!" I worked for a couple of years on that effort, principally looking at memory management tools, and tools that would help users with the massively parallel IBM machines that we have today. Part of my job was to see if we could find out how the users could efficiently run their applications.
In January 2000, I decided, noting that I was turning sixty that year, that I would retire, in June, after thirty-five plus years at the Laboratory. Actually, I was there over a span of thirty-seven, but I left for two years, so I had thirty-five "official years" at the Lab. Now, in retrospect, looking at some of the highlights of my career at LLNL, George Michael had interesting influences my career in several ways. I'm saying that, not just because you are sitting here, but because he introduced me to the Supercomputing Conferences. I can remember when you visited me, George, in my office prior to the first conference in 1988, which you organized with Steve Lundstrom. You and Steve set up the first Supercomputing Conference to be held in Orlando, Florida in response to a need, nationally, for a venue where people who had lots of different experiences in computing, supercomputing in particular, could come together and share their ideas. At the time, you needed someone to put together a program called the "Managers Workshop." The idea behind the workshop was that, in addition to the technical people, there were needs among some computer center managers to have a forum for an exchange of information to learn what's going on currently. I was asked if I would be willing to try to put this workshop together. At that time, it was designed to be a parallel conference with the supercomputing conference. So, we ended up with a program that lasted three days, and I thoroughly enjoyed it.
GAM: It was very good.
JR: It was, and to give George and Steve credit, you had the insight to fill the need of the times: to create a place for people to come to share their ideas. In fact it was "the place to be" for supercomputing. So, I enjoyed it so much in 1988, I stayed on in 1989, and with Bill Buzbee, we did the Computer Center Manager's workshop again in Reno. Then I became part of the team of people who worked on the supercomputing conferences, on and off, over the next twelve years. I attended all the SC conferences from 1988 through 1999. It was a thoroughly wonderful experience for me, because I was able to meet people from other organizations with similar interests. And I think to your credit, George, that you created an entity that has gone beyond all our thoughts about what it would be, resulting in something marvelous. I think it's a great organization. So, that was a highlight, actually I enjoyed all the technical experience I gained at the Laboratory from the 1960s to the 1990s, plus I enjoyed my management experience. I feel that I've had a breadth of exposure to the various aspects of computing.
GAM: Well, lots of people enjoyed your management tenure. It was very good, a breath of fresh air.
JR: Well, thank you. I still remember today, I wasn't in the job a week before I saw that George Michael and Bob Cralle were on my calendar. I spent a couple of hours, with the two of you, learning what should be done. I actually appreciated that kind of input, because I discovered that when you become a middle manager...
GAM: No one talks to you.
JR: Exactly, no one talks to you. It's as if suddenly you've got these "antlers" growing out of your head, so no one talks to you.
GAM: Actually, we did that because Bob Lee invited us to come browbeat him. One of the things we suggested to him was to appoint Mary Zosel as his consultant because he said he didn't know much about computing.
JR: Also, several people have said to me recently, actually before my retirement, "Don't you wish you hadn't left the Lab?" since I returned. Actually, in retrospect, I'm glad I did leave, since the two years experience "outside" was something that was unique. I worked in different groups, and I made professional acquaintances, contacts that I've kept through today. It was an experience I wouldn't trade. I think, George, you went down to North Carolina for a while?
GAM: Oh no, Haverford, Pennsylvania.
JR: OK. For the computing there, and I think it helps to see how other people do things. It's important.
One of the things I was particularly proud to be a part of was the growth in Livermore's computing development. I was not part of the team in the 1950s, but I think that the people that were there then, who took basically nothing and built gold out of it, were a remarkable group of people. When I started in the early 1960s, they had been at it, you and the others, for ten years, and I got to know the people and appreciate the early work. I think they put together a remarkable foundation for a tremendous future.
GAM: It was a lot of very exciting stuff. Fun to work.
JR: I have an enormous respect for Sid Fernbach. He had his style of course, but he built an organization that really produced something. I think that's admirable. I'm just lucky to have been part of that. When I think back of the things that were done at Livermore in the 1960s to make time-sharing a reality based on some ideas out of MIT, the MULTICS system.
GAM: Yes. And Berkeley too.
JR: And Berkeley, and to create it on the hardware that we had, and to make it work, it was pretty remarkable work that the people did in the early Computation organization. I think it was just great! There are so many people at the lab, that I have encountered over the years, who have been a real pleasure to know and to work with. As a retiree, it is the people at the Lab one misses the most.
GAM: I agree with you John. I believe we have come to a finish. I want to thank you for taking the time to discuss your experiences. You had a wonderful career.
