CLINK. CLANK. THINK.
It was the best publicity any executive could ever hope for. For millions of readers, Time equated our products with the advance of civilization. “The prospects for mankind are truly dazzling,” the story said. “Automation of industry will mean new reaches of leisure, new wealth, new dignity for the laboring man.” While Dad kept himself pretty far removed from the computer projects, leaving them to Williams and me, this was the message he had always been interested in putting across. Decades before computers even existed, he had seen that potential in punch-card machines.
The article included a small picture of Dad and told how he had built IBM. Still, I knew I’d feel slighted if I were in his shoes and it were my son getting top billing. That put a damper on my excitement. When I found out I was probably going to be on the cover, I felt obliged to try to cushion the blow to Dad’s ego by sending a letter to him and Mother. “Whatever they say about me is a reflection on you both if it’s favorable,” I wrote. “There would be no IBM for me to be president of had it not been for your initiative and courage, Dad. This I know, and this I have told Time. While I hope I have in me what it takes to do a good job now, I know I don’t have what you had in building the company.” It was a sincere but clumsy attempt to reassure him, though I doubt it made him feel any less upstaged. He didn’t talk to me about the article at all. He never said, “Great going!” and I never brought it up.
By the mid-’50s “computer” was becoming a magic word as popular as vitamins. Top executives rightly believed that the companies of the future were going to be computer run. Board chairmen would say, “We’ve got to get a computer!” Everybody wanted one, even though precisely how to use the machines was still a mystery. It became the conventional wisdom that management ran a bigger risk by waiting to computerize than by taking the plunge.
If Remington Rand had put their money and hearts behind the UNIVAC right at the start, maybe they’d have been in Time magazine instead of us. But nobody at the top of the company had a vision of what computers might mean. Jim Rand was more of a conglomerateur. While Dad was saying “Shoemaker, stick to your last,” Rand’s company was selling everything including office equipment, electric shavers, autopilots, and farm machines. Rand wouldn’t even let Eckert and Mauchly use his punch-card salesmen to market computers—he said it would cost too much. Instead things were set up so that if a new UNIVAC displaced Remington Rand punch-card equipment, the punch-card salesman lost commissions.
At IBM there was never any question—we put the whole weight of our sales force behind our computers as soon as they were announced. At first our salesmen knew almost nothing about them, of course, so we made sure that senior executives and the engineers who did know were available to help them sell. Months before the machines were ready for delivery we hired dozens of graduate mathematicians and physicists and engineers to help customers decide how they might use the computers when the machines arrived. To spread knowledge of the new field, we held seminars in Poughkeepsie for our customers and salesmen both.
In the history of IBM, technological innovation often wasn’t the thing that made us successful. Unhappily there were many times when we came in second. But technology turned out to be less important than sales and distribution methods. Starting with UNIVAC, we consistently outsold people who had better technology because we knew how to put the story before the customer, how to install the machines successfully, and how to hang on to customers once we had them. The secret of our sales approach was the same thing that made Dad so successful in punch cards: systems knowledge. That was where IBM had its monopoly. No competitors ever paid enough attention to it, not even the people at Remington Rand, who should have known better because they were in the punch-card business too.
By the spring of 1954, IBM and UNIVAC were running a close horse race. In terms of computers actually installed, Remington Rand still had the lead by about twenty to fifteen. But our salesmen, racing far ahead of our factories and engineers, had piled up enough orders for us to outdistance Remington Rand four times over. All we had to do now was deliver. Our bestseller was the new computer we’d announced for accounting applications, the 702. We had orders for fifty of these, which we were getting ready to build in a three-year production run starting that fall. The program was on schedule, but so much was riding on the 702 that everybody involved was extremely jumpy. Even Dad felt the tension and worried that other companies were going to steal the business away. “At the rate we are going we will never fill those orders,” he would scold.
Bringing out the 702 on time meant that all of IBM’s departments—product planning, engineering, manufacturing, sales—had to cooperate. I didn’t assume this would happen automatically—the project was complicated, and there were a lot of punch-card men who would be just as happy to see computers disappear. We could easily trip ourselves up, and I decided we needed somebody in charge of making sure that didn’t happen. I chose Vin Learson, who had emerged as one of IBM’s best operating executives. He was six feet six inches tall, and his mere presence in a room was enough to get people’s attention. The job turned out to be one of the most important assignments in IBM history.
By summer our engineers realized, to their horror, that the 702 was probably not the great UNIVAC-beater we thought. One big problem was the machine’s memory. The type of storage circuitry we were using worked faster than the circuitry in the UNIVAC, but it also “forgot” bits of data more often. We could make the 702 perform reliably enough that delivering it to customers wouldn’t hurt IBM’s reputation, but only by providing laboratory-trained teams of specialists to babysit the machines. Our engineers and production managers weren’t sure how to proceed.
Learson turned this quandary into a triumph. His first move was to order a crash redesign of the machine. He took what we’d learned working for the Air Force on SAGE and used it to skip a grade, so to speak, in computer development. The MIT engineers on SAGE had achieved a historic breakthrough in memory technology that involved storing data on arrays of tiny doughnut-shaped magnets called “cores.” Core memory was ultra-reliable, and our engineers had been planning to incorporate it in the next generation of IBM’s computers, about three years down the road. But Vin told them to jump on it right away. He drove the engineers at such a ferocious pace that in less than six months we’d revamped our entire computer line with core memory. Meanwhile Vin decided that we’d go ahead and manufacture the relatively unreliable 702s, but just for a year as a stopgap. As soon as the newer design could be produced, we’d switch our customers to those and either upgrade or replace the old machines.
In a little over a year we started delivering those redesigned computers. They made the UNIVAC obsolete and we soon left Remington Rand in the dust. By the time the presidential elections rolled around in 1956, we had eighty-seven machines in operation and one hundred ninety on order, against forty-one in operation and forty on order for all other computer-makers. Eisenhower beat Stevenson again, but this time the computers you saw on TV were IBM machines.
Whenever we had superior technology to complement our systems knowledge, our business skyrocketed. That happened when we started delivering a small computer called the 650, in 1954. It was far less powerful than the Defense Calculator, but much cheaper. Competitors like Underwood Typewriter and National Cash Register were racing to build small computers that could be used by ordinary businesses, but the 650 outperformed them all. Over the next several years it enabled us to bring thousands of punch-card customers into the computer age. The 650 rented for about four thousand dollars a month, and was the perfect choice for companies eager to try computing because we designed it to work along with ordinary punch-card equipment. Yet it could do accounting jobs that were beyond punch cards. For example, life insurance companies used to spend a lot of money calculating premium bills. Depending on age, sex, and other factors, each life insurance customer was supposed to be charged at a different rate, and typically this calculation was done b
y hand. Clerks would look up the rates in tables and work out the amount due on adding machines. But with the 650, the companies could load their actuarial tables into its memory and the computer did the work. Its ability to handle these bread-and-butter applications made the 650 hot. While our giant, million-dollar 700 series machines got the publicity, the 650 became computing’s Model T.
We played a large role in creating new professions such as programming and systems engineering. As it finally became obvious that we were giving birth to a whole new industry, we discovered that the world wasn’t entirely ready for our machines. It was as though we had the airplanes, but no one to fly them and no place to land. Our customers often complained that the most difficult thing about having a computer was hiring somebody who could run it. They’d ask for help, we couldn’t provide all those technicians ourselves, and there was not a single university with a computer curriculum. Sometimes we even found ourselves in a position where we had to hold back from taking a customer’s order. So I went up to MIT in 1955 and urged them to start training computer scientists. We made a gift of a large computer and the money to run it, and they shared that machine with ten other schools in the Northeast. For the 650, we adapted a very aggressive college discount program that existed for our punch-card machines: you could get 40 percent off for setting up a course in either business data processing or scientific computing, and 60 percent off for setting up courses in both. I put these education policies near the top of the list of IBM’s key moves, because within five years there was a whole new generation of computer scientists who made it possible for the market to boom.
Wherever I traveled during those years I tried to recruit top people for IBM’s research and development side. The engineers who were hardest to attract were those graduating from Stanford and Caltech and the University of California—the smart ones never wanted to leave the West Coast sun to come East. So, very early on, we decided to put a laboratory out in San Jose, and I bought a building that had been intended for a supermarket. The man we sent to manage the new lab was Reynold Johnson, one of Dad’s self-taught inventors from Endicott. He had started out as a high school teacher in Minnesota and he came to IBM in the 1930s proposing a machine that could automatically read and grade multiple-choice tests for schools. Some executives told him the idea was impractical, but Dad overruled them, put Johnson on the payroll, and let him build his machine. IBM made several million dollars on test scoring equipment, and the method is still used on college admission tests today.
Johnson was delighted at the thought of escaping from the rivalries and pressures of Endicott and directing his own lab. He moved to California, hired three dozen young engineers, and in less than three years presented IBM with an invention that was truly spectacular: the computer disk. It stores data in the form of tiny magnetized spots on its surface, and one problem Johnson faced was how to lay down a coating on the disk that was uniform enough to permit this. I remember the day I saw him demonstrate his solution. He stood in front of a spinning aluminum disk with some magnetic coating in a paper cup, and began pouring it slowly, like a milkshake, onto the disk’s center. When the stuff spread out to near the edge, he stopped pouring, and he had a magnetic disk. The machine he invented, which we called the RAMAC, incorporated fifty of those disks stacked like records in a jukebox, except that they all were spinning at once. A little arm would move in and out among the disks, extending a recording head over the disk surfaces to pick up the data that were needed. The descendants of Rey Johnson’s disks are the main data storage devices in virtually every computer system today, from very large mainframes down to ordinary PCs, and they revolutionized the computer’s usefulness. Computer tapes like those used with the Defense Calculator don’t work well in applications where a computer has to look up a particular piece of information—to check a customer’s bank balance, say, or tell how many seats are still available on a particular airline flight. Without Rey Johnson’s disks those applications never would have been practical. To see why, you only need to imagine a music lover who has a collection of both records and tape cassettes. If he wants to play a favorite song on a tape, he has to wait while the tape deck fast-forwards to the proper spot; but with a record, he can move the phonograph needle directly to the right track and hear the music instantly. A computer equipped with a disk homes in on data in much the same way, and the RAMAC made it possible to retrieve information two hundred times as fast as with magnetic tape.
While we were proud of our computers, proud of our disks, and proud of our tape drives, I didn’t fool myself into thinking that IBM had much genuine scientific prowess. We were a maker of electromechanical equipment trying to go into a very sophisticated field with almost no background. Because of this I kept working to increase the flow of technical information into IBM. When we first started building computers, for example, we arranged for John von Neumann, the eminent Princeton University computer theorist, to give seminars to our men at Poughkeepsie. Von Neumann was one of the atom bomb pioneers and he practically defined the modern notion of software; I didn’t understand his work, but I knew how important he was. From then on we kept a steady stream of experts coming in, and we frequently sent our engineers out to university courses as well.
But it was soon obvious that this wasn’t enough, and we began searching for a senior scientist to come and organize a program of pure research within IBM. Wally McDowell, our chief of engineering, spent the better part of 1955 scouring the country for candidates. I took about a month and went around interviewing them, even though I was somewhat in the dark because I didn’t know the scientific community that well. Finally I settled on a candidate who had done an impressive job at building up the school of engineering at a major university. But before I presented my choice to IBM’s board, I attended a meeting at MIT and mentioned his name in a chat with Jim Killian, MIT’s president. Killian looked horrified. He said, “Oh, no, he wouldn’t be appropriate at all for that job.”
“Why not?” I asked. I felt my ears burning because I had apparently done something stupid and didn’t know what it was. Killian was evasive about his reasons, so finally I said, “Look, we want a distinguished scientist. Until a minute ago I thought I’d found one. But if you know of someone better than this fellow, tell me!” What I didn’t understand then, and Killian did, was that American science was dominated by a coterie of men who had worked together during the war. These were people like Leo Szilard and the team that had built the atom bomb, and Isidor Rabi and the team that pioneered radar. The qualifications of the man I’d found were superb, Killian said, but that made no difference. Unless I hired someone from this circle, IBM might spend large sums and still end up with a lab that was merely second rate, because we’d have trouble attracting other top people.
“So who should I call?” I asked, and Killian said promptly, “Emanuel Piore.” He was one of them. I’d never heard of him, but Killian filled me in: until the year before Piore had been chief scientist at the Office of Naval Research, and he had played a major role in funding cold-war military work at universities. Now he’d left the government and was working for a New York defense contractor. I tracked Piore down that night by phone and went to see him the next day. He seemed familiar, and it turned out that his uncle was Michael Romanoff, the self-proclaimed Russian prince who founded Romanoff’s restaurant in Hollywood. The prince had been a celebrity during my nightclubbing phase. Manny looked exactly like him, a tweedy man with bushy eyebrows, shaggy, unkempt hair, a dark complexion, and a perpetual slouch. He smoked a pipe, had a sort of mumbling demeanor, and tended not to look you square in the eye. All the same I found him very appealing and on the strength of Killian’s recommendation I hired him.
Now that the matter was settled, I wondered if I’d picked the right guy. I didn’t know what to make of Manny’s odd, self-effacing manner. But a couple of months later I took him to Zurich to introduce him at the laboratory we had there. Manny was mumbling at the dinner they gave us, and a Swiss
scientist who was pretty arrogant decided he must be a pushover. So he criticized him, in a nasty tone of voice, for not setting sufficiently precise research standards and goals for IBM. Piore came back at him like a lion. “I’ll take that and I’ll answer it,” he said, and gave a detailed five-point response. Then he said, “Is there any facet of this you’d like explained further? I came here tonight to make sure we understand each other. I thought we’d keep things pleasant, but I’m willing to take it right down to the bare knuckles if that’s what you want.” The other guy didn’t breathe another word. I thought, “I’ve got a winner here. Killian was right.”
Piore gave a jolt to some of our product-development engineers as well. They were like sprinters encountering their first marathon runner, and were amazed to see IBM start funding experiments in exotic fields that seemed unlikely to bear fruit for decades, if ever—like superconductors and artificial intelligence. What the engineers thought of as basic research Piore often dismissed as mere long-term product development, and what he called research was so far removed from what the engineers were doing that they saw no reason for it at all. At Piore’s urging we doubled the percentage of our revenues devoted to research and development, and much of the additional spending was earmarked for pure science.
With all this creative ferment, IBM’s potential seemed unlimited as long as we avoided horrible mistakes—a feat much harder than I first assumed. One day near the end of my first year as president, Al Williams came into my office with a very long face. Al was treasurer at that point and he’d just added up the financial results for 1952. “Boy, you’ve got a problem,” he said. Sales that year were up by 25 percent—and profits were barely up at all. I was flabbergasted. Without knowing it, we had allowed our costs to eat up the profits from over sixty million dollars of new business. It was too late to do anything about it. The money—about seventeen million dollars—was gone. Even worse was the way this had happened: we’d spent more than we expected because IBM had no budget. Williams and I had been trying to run this major corporation out of our vest pockets, the way Dad always had. That might have worked before the war, when IBM was a forty-million-dollar company, but by then we were almost ten times that large.
Father, Son & Co. Page 28
