First-Hand:IBM's Evolution, From Punch-Card Machines to High-Speed Computers
Harry D. Young
In August (1945), Selective Service recalled me for another physical and this time I was accepted. I was inducted into the Air Force on September 18, ten days after VJ Day. There were tests to determine the best place for inductees. When interviewed, I expressed my interest in radar work but there were no openings. The interviewer suggested cryptography and I was accepted.
During my basic training, I was pulled out so the service could investigate my background for security clearance. After basic training was completed, I was held up awaiting an opening in the cryptography school. In the meantime, a team had been going around to various bases to test recruits for IBM Customer Engineer School. The IBM opportunity came through first and I was sent to Endicott, New York, for four months to be trained in maintaining various IBM punched card machines.
The IBM school was rather unusual. Army regulations required that we always wore uniforms. IBM insisted that we punch in on a time clock. The rational for this was the insistence that everyone, even the president of IBM, Thomas J. Watson, was on the clock. Classes started at eight a.m. and continued till five p.m. with an hour break for lunch. Our instructor was a former customer engineer from Texas who spoke in a dull, monotonous voice with no inflections. It was so bad that some of the students fell asleep in class. We were taught about the IBM punched card machines of that time. These included the high speed sorter, collator, accumulator/printer and keypunches.
The sorter could take a stack of cards and sort them into eleven stacks, one column at a time. The operator had to replenish the input stack as necessary and empty the output stacks when they became full. When the input stack was finished the operator took one of the sorted stacks and started again with the next column. Sorting a large data base was slow and tedious but considerably faster than doing it by hand. Sometimes the operator would drop the card stack as it was being moved, and the sorting process would have to be completely done all over again.
I went to Greensboro, North Carolina to wait for my assignment. There were many tales of men going through training and then never doing what they were trained for. From there, I went to New Orleans to board a troopship to the Panama Canal. I was assigned to Albrook Field in the Canal Zone.
The IBM installation had previously been in Hawaii. Panama generally used 25 cycle (now Hertz) power but the IBM equipment needed 60 cycle power. The Army ran a power line from another area of Panama for this purpose. When I plugged the keypunch machine in, I noticed an arc from the plug to the socket. There was quite a potential difference between the grounds. It did not seem to bother the equipment but the operators complained about the shocks. To solve the problem, I placed rubber mats under the machines and the operators' chairs to isolate them from ground.
Initially, the equipment was not operating so the IBM office in Panama City sent two men to assist. IBM employees have the reputation of always wearing dark suits with ties. One of the machines had an oil pan at the bottom to catch any oil drips. One of the men lay down on the floor in his suit and tried to get at something in the machine. As he was doing this, his hair dipped into the oil pan and he arose dripping with his new "oily do."
1954. I went to work at the Maryland Electronic Manufacturing Coorporation (MEMCO) in College Park, Maryland, in 1954. I was assigned to Bill Perecinic and worked on various aspects of radar beacons.
1957. MEMCO's strong microwave and antenna group ordered a Luneberg lens for some experimental work. This type of lens is a dielectric equivalent of an optical lens made in the form of a sphere. The dielectric constant varies from minimum at the surface to maximum at the center. When the lens was delivered, the receiving area wanted to make sure it was in the packing box. When they opened it and found a spherical ball of foam material, they thought the lens was packed inside so they cut it apart. Luckily, they were able to use the lens by taping the halves back together.
The microwave/antenna group had a contract with Lockheed to do a radar signature analysis on a new fighter aircraft. When the Lockheed people came to MEMCO from Texas to discuss the contract, they were asked for fuselage drawings to assemble a mock up. Lockheed dissented, indicating the drawings were classified and that it would be too difficult. MEMCO said they could not make any measurements and would not be able to perform on the contract without the drawings. Lockheed finally agreed to supply them.
Several weeks went by without any contact. We finally called and asked about the drawings. The contact said, "Wes sent them out the day after he returned to Texas. Wait a minute while I check my desk. Yes, I have a classified document receipt here signed by your department."
The engineers immediately went out to receiving, searched for the classified drawings and found them. They were loft drawings in the form of four feet by six feet and four feet by eight feet aluminum sheets stamped CONFIDENTIAL.
MEMCO was having a cash flow problem. A small company on the same street in College Park, the Ahrens Instrument Company, had been bought by Litton Industries several years earlier. Our president came in on Monday morning and anounced, "We're going to hold a special stockholders meeting tomorrow night to sell the company to Litton." Since he owned a controlling interest, fifty-one percent, in MEMCO, the stockholders approved the sale and the takeover became effective January 1, 1958.
1947-I was attending the CUNY school of engineering and joined the Institute of Radio Engineers (IRE later becoming the IEEE) as a Student member. I subscribed to the IRE Transactions and recall a report from a mathematician at the Thomas J. Watson Mathematics Lab in Yonkers, New York—a research arm of IBM.
He decried an effort to analyze a rather complicated differential equation. He used IBM equipment consisting of multiplying punches and accumulators and went through several tens of thousand of punched cards.
The multiplying punch would read factors from fields in the card and after multiplying them, punch the result in a third field. This machine was connected to an accumulator that could sum the results and punch the sums in another series of cards. After describing the several months effort in detail, he concluded with the statement, "No longer is it a question that the problem is too complex to solve. Rather the question is, 'Is it sufficiently complex to warrant the effort required to solve it?"'
1953- Seyour Shefter was a mathematician in the Meteorology Branch at Evan Signal Laboratory. He had been going to computer conferences to see how the technology could be used to improve the processing of radiosonde tracking data. He would describe the events at the usual computer conference.
Typically, an engineer from one computer company would state their computer processed at the rate of seventy-five thousand operations per second. Another would describe a computer rate of one hundred twenty-five thousand operations per second and a third would brag of a computer working at the state of the art, i.e. two hundred twenty-five thousand operations per second. Then someone from IBM would say, "Our machine performs ten thousand operations per second for twenty-three hours out of twenty-four and we use the twenty-fourth hour for PM."
All the other people shutted up. Their high speed computers could not keeping working long enough to even approach the performance of the IBM machine.