Computer Memories (continued)
Among other things, I became an expert in the IBM 101 Statistical sorter, and IBM brought people from other users of that equipment, and a couple of times brought me a written definition of a problem someone could not solve, and just for fun, I acted as the expert, and solved those problems. Of course, I became an expert in many other items of IBM equipment.
As time went on, we installed a CPC (Card Programmed Calculator). The CPC was -- in a way -- a stored program computer, with each line of program code stored on a card that was read by an IBM Tabulator. The information was then fed into intermediate storage in a couple of little boxes of memory, each about the size of a small refrigerator, that each stored 480 digits. When calculation was needed, data was sent to and from an IBM calculating unit, with answers being punched in a card punch machine.
I remember at night, as we were going home, they would load a handful of cards into the hopper, and when we arrived in the morning, sometimes the hopper was still not empty. The IBM equipment used in the CPC, were originally called the 604, 415, 513, etc. They were modified and re-numbered for this combination.
We soon installed an IBM 650, -- the first stored program computer that could be ordered from the IBM sales manual and the first I learned to program. The 650 stored its program on a drum, the same storage medium where data was stored. That drum held either 1,000 or 2,000 numerical characters. The program was loaded from punch cards, one instruction per card.
Up until then I had done data reduction of missile test firings, production control for building jet fighters, and we wrote 10,000 paychecks a week, using an IBM 604 with 32 characters of memory. No one could imagine what anyone would ever do with 2,000 characters of memory. A problem that would need 2,000 characters was just out of sight.
Although the 650 was in another part of the IBM Room, the part reserved for engineers, I insisted on attending a programming class at the local IBM office, as my introduction to stored programs.
About this time I attended a presentation given by IBM, about a high-speed computer, called the 701. I remember that the head of production control was sitting next to me, and asked the question, "If these machines are so fast, why do I have to wait so long for my reports?" Of course there was no answer. He was just about 20 years ahead of the industry.
During this time I had borrowed a couple of hundred dollars from my brother to buy our first house, a two bedroom place that cost $7,500. In order to repay the money he loaned us, at night I worked part time at the Lone Star Gas Company, processing the paid-bills through IBM machines; learning and earning as I went.
THE NEEDLESS REPORT, TWICE!
These two stories are almost redundant, but actually happened at IHC in Chicago, then at Chance Vought in Dallas a couple of years later. In both cases, the people in the IBM room had no idea what was going on in the factory, and the people in the factory had no idea what was going on in the IBM room, thus some worthless projects continued for uncounted months. I am sure that my knowledge of both sides of the glass divide enabled me to spot a complete waste of time and effort.
At IHC I had worked on the factory assembly line stocking parts, and repairing diesel engines that failed the test stand operation. My knowledge gained here enabled me to create production control systems for aircraft and missile engines a year or so later.
In the IBM room we had a very large file, 150,000 cards or more, that had to be massaged over and over each week, adding, deleting, sorting, and printing a two-copy report that resulted in two bound volumes, two to three inches thick. No one would listen to the new kid, still wet behind the ears, who tried to tell them about the report, and why it would make no sense on the production floor.
Finally, for three weeks straight I made sure I bound the books backwards, inside out, so that when it was opened, all that was seen was blank paper -- all the numbers were hidden. When I heard no complaints, I came to work early one evening, went to the factory floor to talk to my former boss, the General Foreman. When I asked about the report, he pointed to an adjoining office in which the reports were stacked to the ceiling and said, "I never look at it, and Joe, across the aisle, doesn't use his copy either." And this was a two-copy report.
I had to go over the head of my direct boss in the IBM Department, and finally convinced the big boss of the problem. He investigated, then canceled the project. I was a hero to those who hated that very repetitive job, and a bum to those who now had to learn how to do other things.
At Chance Vought Aircraft, I was able to convince them I was the person to help design the production control system.
While researching I discovered another completely useless project. Each Wednesday afternoon, a huge cart loaded with file drawers filled with IBM cards arrived in the IBM room. The cards were massaged all night and the next day, then returned to the production control office. I found that everyone except those who did the work in the IBM room, and the person who pushed the cart to and fro, completely ignored the card file - it supplied no useful information.
Again, the complaints about my nosiness and complements about my success in having the project ditched were evenly divided. Truly more than one person thought of me as a Yankee troublemaker. They had been doing that Wednesday night job for years, and now I had spoiled what they considered a lifetime career.
RAND CORP. SANTA MONICA, CA
When we moved to Los Angeles in 1955, I got an excellent job at the RAND Corporation, in the IBM Department. One day, soon after I arrived, my boss asked me about doing odds and ends kinds of jobs, a lot of one-time, often times odd jobs. It turns out a couple of people had quit RAND, because they didn't like the variety of jobs assigned to them, and the man who was responsible for those jobs right then, was complaining. I said that if I did a job twice, I wanted it to be because I did it wrong the first time, not just because it was Tuesday again.
Most people I worked with at RAND (outside the IBM Department) had at least one college degree, and some had multiple Ph.Ds. I was told that with my lack of education -- two years of high school plus a GED -- there wasn't much chance of me getting anywhere at RAND.
It always struck me as funny that although I could do the job as needed, and was identified as Dr. Humberd on my airplane tickets and hotel reservations, I could "get nowhere" at RAND. For example, I taught several "Introduction to Computer" classes, and the first formal class I taught at RAND included 17 people with Ph.Ds.
STATISTICAL COMPUTATIONS
I almost became famous in the statistics business with my ability to use IBM card equipment to compute bi-serials, chi-squares, product-moment correlations, and other things, before the big computers could be programmed to do these jobs. They told me I was the first to be able to solve those problems using IBM cards and the IBM 607 calculator, with, if I remember correctly, 56 characters of memory.
Several times while someone with a Ph.D. with a big statistical contract to complete, or someone who had to complete a set of calculations for his doctoral dissertation was being interviewed for employment at RAND, I was called in to see if I could understand their computation needs well enough so I could do the computing and let the poor applicant get on to something more useful. They always got a big kick out of my lack of formal education.
In more than one case I was mentioned in the dissertation as the one who helped with the calculations. I will always remember when one man opened the package containing his Ph.D. diploma from Harvard, and saw it was printed in Latin, he exclaimed, "All that work, and I can't even read it!"
TRAINING THE AIR DEFENSE COMMAND
RAND Corporation had a contract to train the people who manned the Air Defense System. This project had gotten its start in a fascinating manner. People at RAND had wanted to study the interface/reaction between a man and a machine. At first, an IBM Room with the machines and the operators provided the best environment they could imagine. Boards had to be wired, cards had to be punched, then fed into this machine and that, a human worked with the results of one operation, then had to follow the next procedural step, and on and on, to produce the final product.
As the next stage of this study, they analyzed the interface/reaction between the operators of Air Defense RADAR screens, and the Air Defense System. When they had completed that analysis, it was discovered that the men who had been exposed to the RAND experiment were the best trained men in the Air Defense system, so RAND received a contract to train the rest of them.
At first, training maps were plotted by a human, then cards were punched that when read into an IBM 407 Tabulator, would result in multiple copies of a printed page with a RADAR screen background, with airplane identifications and locations printed a little further along their flight path, then were shown on the previous map. Those maps were read, pages were turned, a script was followed, as the rather crude training system advanced. It was slow beyond imagination, and not easily applied to a wide variety of RADAR site locations with a wide variety of flight patterns.
When the IBM 701 Defense Calculator became available, programs were written that would calculate airplane flight patterns, display them on a special CRT where they were filmed by a 70 MM camera. In a real Air Defense RADAR Installation, the film was then fed into a device that would display the flight patterns on the RADAR screens. Then, using the first real Video Game, another machine (whose number I have forgotten), would place a blip on the screen, and by verbal commands from the man concentrating on the screen, an operator could direct the blip, representing a fighter plane, here and there across the RADAR screen, until the "unknown" was either identified or "shot down."
At first, the 701 had 4096 36 bit words of electrostatic tube memory. You could go in back of the memory boxes, and in the dark you could see the bits that were lit and those that weren't, in a 36 bit memory word.
It might be nice to consider that our country was being protected by men reading a RADAR screen and giving directions to other men standing behind a two story transparent plastic screen etched with a map, who were using wax crayons to write numbers and codes backwards, so they could be read by officers at the front of the screen, who made decisions concerning "friend or foe." But we survived!