Enable JavaScript.

1974: Systems Research Institute


Hal Lorin had been trying to quit smoking all through the ten week term of our class, but he wasn't having much success. He'd get about half way through a ninety minute class, and then suddenly, in mid-sentence, he'd reach into his pocket and pull out whatever loose change he happened to have. This, he plunked down on the desk of a student in the front row as he grabbed the pack of cigarettes lying there. Hal did this almost daily, in our Operating Systems class at IBM's Systems Research Institute. The last day of the class was no different, but this time his eyes went wide with surprise. There, among the perfect, machine rolled cigarettes that came from the pack was one that was unmistakably hand-made and of clearly questionable content.

Carefully, as the class howled with laughter, Hal pushed all of the cigarettes back into the pack, returned the pack to the student's desk, and retrieved his pile of loose change. "This does it." he said. "I have smoked my last cigarette."

IBM always spent lavishly on employee education, and for technical people, the ultimate was a term at SRI. Targeted at employees with about five years of experience, SRI offered a wide curriculum of courses related to computer systems, with human relations and leadership courses added for good measure. Its faculty was a mixture of IBM professionals and professors from New York Universities, all experts in their fields and wonderful teachers, besides. SRI was located on East 42nd Street, near Grand Central Terminal, in New York City, and the students were housed in the old but serviceable McAlpen Hotel, a few blocks away, near Macy's. The students were all adults, hungry to learn, but there was always someone organizing an excursion to a theater, a concert, an art museum, or just to dinner somewhere in the city. IBM's generous per diem made a lot of fun possible, and helped to build terrific camaraderie.

Altogether, SRI was a unique opportunity, one that I had been longing to experience. I had my chance in the Fall of 1974, after finishing my part of the Mohawk project. Fortunately, Endicott isn't very far from New York, so I was able to get home every couple of weeks. Lois was finishing her Masters degree in Special Education at SUNY Binghamton, and was very busy. It was important that I get home to spend some time with our sons, now five and two.

The day I checked in at SRI, I was given a packet of information that included the names of all of the students in our class. Among them, to my delight, was that of my friend Bob Buhlmann. We hadn't kept in touch after I had moved to Endicott, and it was pure chance that we ended up in the same class at SRI. This was great, it gave me a buddy to hang around with.

The real gems of SRI were its faculty. Joe Flanagan taught a course in Queuing Theory that sparkled with his insight and wit. I'm not sure I completely understood what Semi-Markov Chains were all about, but if I ever wanted to model a queuing system, I knew who to talk to. I believe it was Joe who originated the phrase "Open the Kamona." He used it frequently to describe the process of examining the contents of a queue. He also had a habit of pantomiming this phrase by pretending to open a robe and flash the class. In any case, for many years "opening the kamona" was a common IBM idiom, used whenever something needed to be examined closely.

James Martin taught courses in telecommunications. He had a special classroom set up for his use with stereo sound, five screens in an arc at the front of the room, and a variety of slide and movie projectors in the back. He would pace back and forth as he lectured, with a hand-made control panel in his hands that he manipulated to present a collage of sounds and images. The only thing I have ever seen that could compare with James Martin's classes were at Disneyworld or at a World's Fair, and they were all computer controlled. Martin did it by hand. I wouldn't have missed one of his classes for anything.

There was also a professor from New York University, whose name I can't remember, who came in to teach a course in Operations Research. This field was developed during World War II as a way of optimizing the efforts of the Allied Forces. After the war, it was applied to many civilian problems and eventually led to the development of the field of Management Information Systems. Our professor had a thousand stories to illustrate his points. My favorite had to do with the Holland and Lincoln Tunnels under the Hudson River.

Each tube of these tunnels has two lanes of traffic going in the same direction. But the number of cars that could travel through one tunnel was for some strange reason less than the number that could travel through the other tunnel. Using Operations Research techniques, it was finally determined that the difference had to do with the way in which tiles had been applied to the walls of these tunnels. In one, the tiles were applied parallel to the roadbed, while in the other they were applied parallel to the riverbed. This difference affected the speed at which cars typically traveled. The tiles parallel to the riverbed gave drivers visual clues that they were descending or ascending, and they adjusted their speeds accordingly. In the tunnel without these visual clues, cars just kept up with the traffic. The simple solution was to paint lines on the walls parallel to the roadbed to eliminate the visual clues.

Bill Smith handled the soft side of SRI. His classes in Human Relations were extremely popular. His message was that we were human beings with needs and drives that went beyond the workplace. To be productive at work we needed to live satisfying lives outside of work, as well. This was an integral part of IBM's management philosophy It was expressed and practiced in many ways, from the IBM Clubs at most sites, to IBM Country Clubs in some of the older sites, and Family Picnics in many others.

Charlie Bontempo was SRI's database guru. He taught the principles of hierarchical and network databases, but he was a particularly strong advocate of the new relational databases invented in IBM's Almaden, CA research lab.. Charlie also had strong a interest in the concept of distributed databases The importance of people like Charlie in spreading the word about new technologies throughout the length and breadth of IBM couldn't be overestimated.

Years later, when I was working on the distributed files part of IBM's Distributed Data Management architecture, John Bondy had me call Charlie to get his opinion of what we were doing. He said that if I wasn't working on distributed relational databases, I was wasting my time. I couldn't give this opinion much credence since my customers specifically wanted an architecture for distributed files. In retrospect, though, he was right.

As a programmer, one of my favorite teachers was Glen Meyers who was doing terribly interesting work on program composition while teaching it at SRI. He theorized that in a large programming system, there were varying degrees in which modules could be independent from each other and varying strengths by which modules could be bound to each other. He postulated that in an ideal system, modules would have maximum independence with minimum strength bonds to other modules. Glen's book contained numerous examples of composition derived from studies of large IBM systems. From today's view point, Glen was an early advocate of what we now call object encapsulation, the concept that an item of data should be encapsulated by a limited set of modules that alone are allowed to access or operate on that data. I would later see this concept partially realized in the System/38, and fully realized in the Smalltalk programming language.

And then there was Hal Lorin, who taught the Operating Systems courses. The author of several books on the subject, his lectures were primarily chalk and talk affairs. Illustrating concepts on the board with his right hand as he spoke, he tended to erase the board with his left shoulder, making only occasional use of erasers. At the end of each class, the left side of his blue suits were a chalky mess. His dry cleaning bills must have been out of sight.

Hal became an occasional mentor for me. I asked him once why someone with his range of knowledge and experience was teaching, instead of being in the labs developing new products. He said "I've been there, and fought the battles. I see what you guys go through. I just don't have the fight left in me." Regardless of what he said, Hal continued to be a scraper. He had just changed battlefields, and was now fighting for the hearts and minds of IBM's up and coming young operating system designers.

On another occasion, when I interviewed for a job at SRI, I told Hal that I had my choice of jobs, that I could work for SRI, for Research, or in the mainframe architecture department in Poughkeepsie, NY. The Research job had a tremendous appeal to me because of the possibility of eventually exploring ideas relating to object-oriented programming that intrigued me. But in the architecture job I would have a more immediate impact because of my deep understanding of the System/38. The concepts of uniform virtual memories were, then, just beginning to be considered in Poughkeepsie.

Hal explained to me that I had to choose not just between jobs, but more importantly between being a professional and being a company loyalist, and that professionals always hold loyalty to the ideals of their profession above loyalty to any particular employer. Only in this way could you expect to do your best work and thereby give your employer full measure. He expressed contempt for the company loyalists who, he believed, could only lead the company astray. This was a new concept to me. Up to then, I had seen myself as an IBMer who worked as a programmer. But now, I was forced to reassess this position, and I concluded that he was right. I took the job in Research.

In the mid 80's, as a cost savings, IBM consolidated many of its employee education programs, including SRI, and constructed an education center in Thornwood, NY, just north of White Plains. The new campus, set on a wooded hill, included all of the amenities one would expect of a small college campus: comfortable classrooms, a TV studio, conference rooms, a large auditorium, a gym, dormitories, and a cafeteria. The SRI faculty was transferred to Thornwood, but the ten-week SRI course soon ceased to exist. Many of the same courses were taught, but individually or as part of more specific three week programs.

Several times, Hal asked me to come to Thornwood to lecture on distributed data, and on one occasion, I stayed overnight in the Thornwood dormitory. I then understood why the ten week program had been discontinued. What a boring place to spend a night. I couldn't imagine spending ten weeks there. More importantly, though, one of IBM's finest assets had been destroyed in the name of false economies.


1974: L'institut de recherche sur systèmes


Désolé, pas encore traduit.