The following is an excerpt from Digital Equipment co-founder Harlan Anderson's autobiography "Learn, Earn and Return," courtesy of Locust Press.
Read Network World's interview with Harlan Anderson.
It was July of 1952, and I had accepted MIT’s offer to come to work in their Digital Computer Lab, which at the time was part of the institute's Department of Electrical Engineering. It was widely known at the time that the lab was involved in the development of the Whirlwind Computer. Whirlwind, one of the first stored program digital computers, was similar to the ordvac—which I became familiar with while at the University of Illinois.
But in addition to Whirlwind, there was a lot happening at the Digital Computer Lab I was not told about because of military security. This was not the first time the U.S. military had come to MIT; there had earlier been an extensive and successful collaboration between MIT and the U.S. Department of Defense during World War II.
One of these collaborative projects included the development of radar, which was a very successful joint effort between U.S. and U.K. scientists. MIT’s part in this collaboration was known as the Radiation Laboratory. Technical projects like this and those undertaken by other institutions, such as the development of the atomic bomb and the technical intelligence work that broke the German Enigma coding technique, demonstrated that science and technology were extremely important to military operations.
In short, the military appreciated the long-term role universities could play, and was willing to financially support them—even during peace time. MIT also benefited greatly from this relationship. The Office of Naval Research had paid for the development of the Whirlwind computer project to explore digital computers as an alternative to analog computers in simulating aircraft performance. Whirlwind was soon to take on a totally different role than that for which it had been created. This was a powerful testimonial to the versatility of computers.
This was the environment I was entering with my first engineering job. The initial phase of this ongoing venture with the military involved a “Project Charles” study to see if a computer based command and control tool could be developed to provide military personnel with real-time, comprehensive pictures of enemy aircraft based on information from multiple radar sets at various locations. This study had been going on for some time before I arrived. When it was successfully completed, a decision was made to implement a small scale version of this concept using the Whirlwind computer as the center of this system. This was known as the Cape Cod System.
The Cape Cod System involved the incorporation of existing radar sites in the general vicinity, as well as establishing digital communication lines to relay the radar information to Whirlwind, which was located in the Barta Building on the MIT campus. Computer programs were then written to convert the information into maps with aircraft locations that were presented to military operational personnel, who could then make decisions about what action should be taken against “enemy aircraft.” Such action included dispatching authentic military fighter planes from nearby Hanscom Field and other military bases to intercept real bombers flown by the Strategic Air Command acting as enemy aircraft set to attack the East Coast. The computer calculated the instructions to guide the fighter planes to a successful interception of the “enemy.”
This very real demonstration probably dazzled the military leadership, because not long afterward, in 1950, the decision was made to make this a full-scale partnership managed by MIT, called Project Lincoln. A few MIT professors held key jobs in Lincoln Lab, and so the Digital Computer Lab soon after became known as Division 6 of Lincoln Lab. Most of its personnel were engineers like me who had been recruited for this new laboratory.
Other divisions were developing communication tools and converting radar site information to digital format from analog. One of the weakest links in this whole system was the Whirlwind computer memory, which was comprised of storage tubes—the most common type of memory in most early computers. It was agonizingly slow and unreliable.
Jay Forrester had been developing a new kind of computer memory called a coincident current magnetic core memory as an alternative to these storage tubes. When I first met him at Illinois on a Sunday morning in 1951, he was on his way to a physics society meeting in Chicago to deliver his famous report on his invention.
The memory’s early tests had been very encouraging, but at the time it was considered too risky to replace the storage tube memory in Whirlwind, which was now critically involved as the heart of the experimental Cape Cod System. The feeling was that the only good test of the new memory would be to install it in a real computer and see if it worked. The only problem was that there wasn’t another computer available for this purpose.
The decision was made to construct an entirely new computer for the sole purpose of testing Forrester’s new memory. This magnetic core memory was a genuine nightmare to assemble, since each bit of information had its own tiny magnetic core which required four wires running through it. A typical small size memory would have more than 16,000 of these cores.
The job of designing and building the Memory Test Computer was assigned to Ken Olsen, a young MIT graduate. My first job at Lincoln Lab was as an engineering member of that project. It was an exciting time for me, and Ken Olsen was a dynamic leader of this
crash project.
As I was to discover, there were several smaller projects being undertaken all around me that would eventually coalesce into a much larger one. When I first joined MIT, I had no idea of the big picture that I just described, due partly to security considerations. It was more than five years before the military went public with details of the new air defense system that MIT was developing for them.
During this same time, Lois had also found a job at MIT that utilized her psychological training. But it was some time before we learned that she too was going to be working at Lincoln Lab. Her job there involved how to best design the man-machine interface of this massive new system—including how the operator’s console could be optimized, how much information to show on the screen, the size of the letters and symbols, the shape of knobs, etc.
This was a semi-automatic system with military personnel making the ultimate decisions based on the graphic pictures the computer presented to the operator(s). A light pen had been invented so that an operator could point at something on the screen and cause some action to occur for that aircraft. This was the forerunner of the computer mouse in common use today.
There was no place available on the MIT campus for the emerging Lincoln Laboratory. Instead, the university had rented temporary space in buildings near the campus to facilitate the project’s early stages. My first office was in the former Whittemore Shoe Polish factory, which was one block from the campus near Massachusetts Avenue.
Lois and I had first driven to Cambridge in our very old car, which was crammed with our personal possessions plus Lois’ younger brother Ken, who was seeing the East Coast for the first time. The few things that didn’t fit in the car were being shipped by moving van. Our first apartment was in a home in nearby Belmont that was not zoned for multiple families. The landlord lived downstairs and we had to park our car on the street. Public transportation was also close by.
That first July was hot, and having lived our entire lives in the Midwest, we sought a respite from the heat at nearby Revere Beach. We were not familiar with the risk of sunburn at the beach and unfortunately stayed much too long. We were unable to go to work for several days due to overexposure. Belmont was our home for almost a year before a significant change in my career loomed over me.
The job at Lincoln was ideal. It was deeply immersed in advanced computer development. The lab was streaming with visitors from all over the country and from England. They were curious about new concepts in computer applications—namely real-time use of computers with man-machine interactions. Many of these scientists were from other government sponsored labs, but they never told us about why they were interested in computers due to security considerations surrounding their projects—which presumably were unrelated to our work.
As I mentioned previously, English engineers had helped to break the German cryptographic system of sending messages to the troops in the field. Within the last few years, I have visited Bletchley Park, located north of London, where that secret work was performed. I became fascinated with it as a hobby and have since built a working replica of the German Enigma encoding device and given lectures on how the code was broken.
Our memory test computer, known as MTC, consisted mainly of vacuum tubes. The control part was comprised of modules from the Burroughs Corporation normally used as digital test equipment interconnected by coaxial cables with hand-twisted connectors known as BNC (Bayonet Neill-Concelman) connectors. There were hundreds of these modules that had to be interconnected to make up the control portion of the computer. Each module was 3.5 inches high and 19 inches wide. This turned out to be a physically demanding task where we had to wear gloves to keep from cutting our fingers from the many connections that had to be hand-screwed together.
Unfortunately, at the time, there were no test programs available for this one-of-a-kind computer. Wes Clark, one of the most intelligent computer engineers I’ve ever known, came to the rescue. He wrote a very short program called the Inch Worm. The sole purpose of this program, which was stored in the memory, was to pick itself up, move to a new section of the memory and run itself again in the new memory location. If it made its way successfully through the entire memory, this meant all locations of memory were working correctly.
The possibility of having one bad memory location could not be tolerated.
Another concern was that maybe a memory location would work only for a period of time and somehow “wear out”. The inch worm program was run successfully for hours and days at a time, and the excitement it created in the whole lab was overwhelming.
The memory was small by today’s standards—only 1,024 locations— but it worked! Wes had also programmed MTC to play “Yankee Doodle Dandy” as a bonus. This was probably one of the first times a computer had ever played music. It was a nice touch to the successful completion of a crash program to build a computer. MTC is frequently mentioned in computer history literature and books.
Only a short time later the core memory was moved to Whirlwind, and a new memory was made for MTC. At the time, the cost of this type of memory was reported to be $1 per bit. Today you can buy millions of bits for $1.
In 1953, Lois and I bought our first new car—a Chevrolet—with borrowed money from the Harvard Trust Company. Later on, Ken Olsen was interested in buying our very old car, which had lots of problems. I’ll admit I was very reluctant to sell it to him since he was my boss. I insisted that he have a mechanic look it over before I would sell it to him. He bought it anyway. In that same year, my father died of congestive heart failure at age sixty-six. Mother lived alone for the next thirty-one years.
THE CONFLICT OVER ME
When I joined MIT, I notified Selective Service Board #200 in Freeport that I was no longer a student. The Korean War was on and young men my age were being drafted. At Lincoln Lab there were quite a few young military officers working side-by-side with me.
When it was sensed that the Illinois Draft Board was planning to draft all able-bodied men, MIT attempted to get a deferment for me, but they were probably unable to make a convincing case to the people in Illinois because of military security restrictions on what I was actually doing. When that strategy failed, they attempted to get me a direct commission in the Air Force, and have me assigned to Lincoln Lab.
The bureaucracy to accomplish this was time consuming, and Illinois told me to report for duty as a draftee in Illinois. This was delayed by having my point of induction transferred to Boston.
Eventually, time ran out and I was told to report for active duty at 8:15 a.m. on June 19th, 1953 at the Boston Army Base. Lois and I went out to a final dinner together the night before and we were reluctantly anticipating living apart for the first time in our young married lives. However, when we returned to our apartment in Belmont, there was a telegram waiting for us saying that Colonel Armstrong, head of the entire selective service system for Illinois had ordered the Freeport Draft Board to issue a thirty-day postponement of my induction.
Shortly after that, General Hershey, then head of the Selective Service System for the entire country, personally ordered a postponement for an indefinite time period.