The Chip: How Two Americans Invented the Microchip and Launched a Revolution

by T. R. Reid

  There was a strong streak of irony in this reaction. As the Japanese knew perfectly well, their quality mechanisms were hardly secret, particularly to Americans. After all, Japan had learned virtually all it knew about quality control from an American. The teacher was a federal bureaucrat, an obscure fellow, unknown in his homeland but recognized and respected throughout the Japanese islands. He was, like Jack Kilby and Robert Noyce, a son of the prairie.

  W. Edwards Deming was born in Sioux City, Iowa, in 1900. Growing up on the prairie, young Deming, like many intellectually gifted youngsters, displayed talent both in mathematics and music. Eventually, the lure of math and science prevailed, and as an undergraduate at the University of Wyoming, Deming concentrated on math, physics, and engineering. He went on to graduate work at the Colorado School of Mines and at Yale, where he took a Ph.D. in physics in 1928. Dr. Deming then began a long career in federal service, first as a scientist in the U.S. Department of Agriculture’s Bureau of Soils and later, when statistics became his dominant professional interest, at the Census Bureau. Eventually he left the federal service and began an extraordinarily vigorous lecturing and consulting business that he ran from the cluttered basement of his Washington home.

  When I first met Dr. Deming he was past his eightieth birthday and traveling on business about 300 days per year, to Europe, Asia, and various corners of the United States. He was a tall, crusty, imposing gentleman with a fluffy white crew cut, spectacles, and a jutting nose that gave him the look of a tortoise poking around outside the shell. He told me that he had developed a method of squeezing two workdays out of one: he would work until midafternoon, nap until eight-thirty at night, and then rise for several more hours of business. He found time, naturally, for music—sessions at the piano with his wife, Lola Shupe Deming, and their children and grandchildren. He composed liturgical music; the Deming oeuvre includes two masses and a number of canticles. Otherwise, he was all business. I asked him once—he was eighty-three at the time—if he ever took a vacation. “Well, I’m not doing anything for the next few hours,” came the gruff reply.

  A nine-page “list of principal papers” that Deming provided his consulting clients reflects the changing focus of his interests over the years. The first dozen or so of his publications—many signed jointly by W. Edwards Deming and Lola Shupe—deal with physics and chemistry: “Equipotential surface electrons as an explanation of the packing effect”; “Note on the heat capacity of gases at low pressure.” But midway through the 1930s the titles begin to reveal a developing interest in probability and statistics: “On the frequency interpretation of inverse probability”; “On a least squares adjustment of a sampled frequency table when the expected marginal totals are known”; “On the elimination of unknown ages in the 1940 population census.”

  This shift in Deming’s scholarly output resulted from his growing appreciation for the enormous power of statistics to explain precisely what was going on in any repetitive process, including mass production of commercial goods. Deming learned that the statistician could act as a skilled detective when things went wrong in an industrial operation, searching his data to pinpoint the problem and eliminate it. He put this idea on paper in a seminal monograph, written in 1934 and updated thereafter, titled “On the Statistical Theory of Errors.” The “Theory of Errors” was Deming’s explication of the ideas of Dr. Walter Shewhart, the American mathematician who first applied statistical methods to the control of factory operations. Deming set out to be the evangelist of the Shewhart gospel, which held that careful statistical records were the essential key to consistent quality in manufacturing.

  In the lectures he delivered around the world, and in his basic text, Quality, Productivity, and Competitive Position, Deming kept on preaching the lessons he had set forth back in 1934 in the “Theory of Errors.” He stressed that quality and productivity result from diligent observance of certain fundamental rules—he has gathered them into a table of “14 points”—all of which can be summarized in a single, golden rule that should govern all work: “Do it right the first time.”

  Reduced to those six words, the Deming message seems obvious—a point the professor never failed to drive home. “It’s so simple,” he said in his lectures. “It’s so obvious.” But to carry out the obvious required detailed effort. To achieve consistent quality, those involved in any operation, from manufacturing semiconductors to managing a baseball team, must maintain “statistical control”—that is, careful, regular measurement of all aspects of the job. “By describing statistically exactly what is done,” Deming said, “the method locates your problems and leads to innovations that solve them.” When a manufacturing problem occurs, the quality control officer should probe the statistics like Sherlock Holmes probing the body at a murder scene.

  One time, for example, Deming was retained by a shoe company that had run into costly but inexplicable manufacturing delays. Management was baffled; the same work force in the same factory had suddenly fallen far behind its normal production rate. Poring over time cards, maintenance charts, purchasing records, and the like, Deming found that someone had recently ordered a new brand of thread. Now it was all so obvious. The new, cheaper thread kept breaking, forcing workers to stop and rethread their sewing machines time and again. “To save 15 cents per spool they were losing $150 per hour rethreading the stuff,” he explained. Better thread was purchased and the problem disappeared.

  Deming argued pointedly that statistical control is the only sensible means of quality control. “All these companies are running advertisements on the TV about their rigorous inspectors,” he snorted. “Inspection is too late, don’t you see? It’s so obvious. By the time the product gets to an inspector, the quality, either good or bad, is already in. Do you want to burn the toast and scrape it, burn the toast and scrape it—or do you want to make the toast right before it gets to your inspectors?”

  The right way to make toast, under the Deming rules, would involve careful observation of every element of the process—the quality of the bread in the market, the level of light in the kitchen, the wiring and the timer in the toaster, the methods of inserting the bread into the slot and of taking the finished toast out. If all these variables were tracked on statistical control charts, a manager could see just what had happened when any piece was burned. The problem could be corrected quickly. The control process would be more efficient and better for workers’ morale than inspecting and scraping every piece of toast.

  Moreover—and this is the Deming route to enhanced productivity—statistical control is cheaper. It will always cost less to make one piece of toast right than to burn one piece, inspect it, reject it, and then make another piece the right way. “The total cost to produce and dispose of a defective item exceeds the cost to produce a good one,” the professor’s text says. The time and money previously spent to inspect and scrape burnt toast can be directed to a more productive goal—making more toast—once a predictable level of toast-making quality is achieved.

  As the challenge from Japan and other foreign competitors became acute, late in the twentieth century, American industry learned the hard way that the Deming rules should be heeded. In the 1930s, though, things were sharply different; when W. E. Deming talked, nobody listened. There was a brief spurt of interest in statistical quality control at the start of World War II, when quality of production became a matter of national survival. Engineers from munitions plants were brought in to learn from Professor Deming, and the War Department gave courses in factories. “Brilliant applications burned, sputtered, fizzled, and died out,” Deming wrote later in his typically acerbic fashion. “Quality control departments sprouted. They plotted charts, looked at them, and filed them. They took quality control away from everybody else, which was of course entirely wrong, as quality control is everyone’s job.”

  Deming was not a man to give up easily, but the wartime experience led him to focus on other applications of statistics, including demographics. In the global reconst
ruction at the end of the war, he became a sort of roving ambassador of demographics. He was an official observer at the Greek elections in 1946, and from there went to Delhi to advise on the Indian census. Eventually he was summoned to Japan by General Douglas MacArthur during the postwar American occupation to assist in various population and housing studies. And there, at long last, Deming found an audience that cared about quality control.

  A Japanese statistician, Dr. E. E. Nishibori, who was familiar with the “Theory of Errors,” discovered more or less by accident that the author of that insightful paper was in Japan. Nishibori tracked Deming down: would the American be interested in speaking at a quality control workshop for the Union of Japanese Science and Engineering? He would. He did. Another workshop followed, and another. Someone else arranged for Deming to speak in Tokyo at the Industry Club of Japan, a business round-table whose membership included senior management of every major Japanese manufacturing concern. In that speech, on July 26, 1950, Deming declared, to the astonishment of his Japanese audience, that Japanese quality would soon be the best in the world.

  “I predicted,” Deming recalled with great relish three decades later, “that Japanese manufacturers would come to dominate world markets and have their competitors crying for protection. At that time I was the only person in the world who believed that. . . . But it was so simple. You could see that this society was receptive to the ideas that are necessary for quality and productivity.”

  That speech turned out to be decisive. “Once you convince senior management that this will make a difference, the hardest job is done,” Deming always said, and Japan was Exhibit A for him. The Japanese were so convinced by what they heard that they invited Deming back to Tokyo year after year long after the occupation had ended to spread his gospel. The Union of Japanese Science and Engineering distributed millions of copies of his books and pamphlets. Tens of thousands of Japanese factory workers, engineers, and executives studied his teachings at regular classes (they still do so today). These diligent students put his ideas into practice, with consequences that shook the world.

  When Deming arrived in the shattered remnants of postwar Tokyo, Japanese goods were everywhere considered cheap, shoddy products made from cheap, shoddy materials. By the 1970s, the once-proud American automobile industry bought television advertising to boast that its products were made the Japanese way. The label “Made in Japan” came to represent the world standard of quality for products ranging from steel, automobiles, and heavy machinery to cameras, scientific instruments, and consumer electronic gear.

  There is no single explanation for Japan’s postwar economic miracle, but the Japanese, at least, have given a good deal of the credit to their American teacher. Deming was one of the few foreigners ever to receive the Second Order of the Sacred Treasure, Japan’s premier imperial honor. His name and profile adorn the Demingu Shō (the Deming Prize), an annual industrial award carrying the stature of the American Pulitzer Prizes, which is awarded, Oscar style, in an annual presentation telecast live to the entire nation. The name “Demingu” has become virtually a household word.

  And then, fifty years late, the name “Deming” came to have meaning in other countries as well—even in the United States. The Japanese and other competitors led American management to pay tardy heed to the native-born Cassandra; if it were not so, Deming would not have had to squeeze two workdays into one. Sometime between his seventy-fifth and eightieth birthdays, Deming emerged as the guru of what was called the Third Wave of industrialization, following the First Wave at the start of the Industrial Revolution and the Second Wave, which swept in with techniques of mass production. The Third Wave, suited to a world with higher standards and a clearer recognition of limits, focuses on the most efficient possible uses of resources to produce goods of predictable reliability and quality.

  Dr. Deming loved to remind people that he saw the new wave coming long before his countrymen caught on. “The way to compete in an international economy is to promote efficiency, to produce better quality than the other people are producing,” he would say. “It’s so simple. It’s so obvious.”

  The point was somewhat less than obvious to the American semiconductor industry until the startling Japanese success in memory chips brought the message home quite clearly. The Americans took it to heart. The semiconductor firms made quality control—not just inspection but genuine control—a high-priority challenge. It made a difference. In an interview two years after he released his bombshell, Richard Anderson, the Hewlett-Packard executive, said his firm had found marked improvement in the quality of memory chips coming from American suppliers; in fact, he said, the Americans had matched the Japanese on quality standards. This change helped U.S. firms take back a small share of the world market for 64K random-access memory chips from the Japanese.

  “U.S. manufacturers, until the advent of Japanese competition over quality, had made a tacit decision that fast, volume output with component testing to cover imperfections in the manufacturing process was more important than high quality,” reported a study published by Congress in 1982. “The Japanese instead concentrated on perfecting their production process to deliver higher quality devices. As U.S. firms retool and expand capacity, they have apparently been ‘tweaking’ their production process to deliver higher quality devices. It may well be, then, that the Japanese ability to use quality as a penetration strategy will not carry over to the next round of competition.”

  But the Americans did not “tweak” their quality standards enough to head off the Japanese challenge. By the mid-1980s, the sun seemed to be setting on Silicon Valley. Japan’s semiconductor giants had become the dominant force in the world semiconductor market. Despite the inspiring presence of the two inventors of the integrated circuit in its midst, the United States fell further and further behind in overall market share. The Japanese lead was greatest in the field of memory chips, a product that played precisely to Japanese strengths for dependable, high-quality production of a tested product. But firms like Hitachi, Fujitsu, and Nippon Denki also began to take the market lead in certain logic chips and analog chips—product areas where the United States had always been dominant. For a while, Nippon Denki (NEC) even competed with Intel in the microprocessor business for personal computers, until the threat of a patent infringement lawsuit led the Japanese to drop out of that market.

  The annual market share surveys put out by Silicon Valley research firms like DataQuest and VLSI Research, announced in January each year, became a somber event for American chip makers. Year after year, Japan had a clear lead in global market share for semiconductors overall, and in most distinct market segments. Beyond that, Japanese machine-tool makers pulled ahead of the Americans in the market for semiconductor fabricating machines. The lead in fabricating machinery was, in a way, even more disturbing. If Japanese machine-tool makers gave their Japanese customers first call on each new model, then the Japanese chip companies would have another advantage over their U.S. competition.

  The Semiconductor Industry Association, once the proud voice of a global leader, now began to sound very much like the steelmakers and carmakers and television makers that had earlier succumbed to Japanese competition. The SIA began badgering the government for help to stave off the Asian invaders.

  To deliver its warning that the Japanese were coming, the men of Silicon Valley selected one of their own—a senior statesman of the industry who seemed perfectly suited to the role. The designated spokesman was articulate, intelligent, and immediately impressive. He complained that the Japanese were cheating, using their monopoly in the closed domestic market to fund predatory pricing in foreign markets. “Many of the Japanese practices are practices that we would see as unfair, illegal, or whatever,” the spokesman told a congressional committee. “. . . It may be that we can hope the Japanese will play by the rules of our game, but I don’t see any motivation for them to do so whatsoever, since they perceive that they are winning using the current strategy. And, inde
ed, they may win.”

  The spokesman who offered this mournful plaint presented the same plea for government help week after week at hearings, seminars, conventions, and press conferences all over the country. He performed the job as if he had been doing it all his life, but he had not. The spokesman had been a physicist, then an inventor, then a corporate manager and a venture capitalist—and had performed all those jobs with equal facility. The spokesman was Robert N. Noyce.

  In the first edition of this book, this chapter came to an end at that unhappy juncture for the American semiconductor industry. Bob Noyce’s emergence as spokesman for the industry came at a time of desperation among semiconductor firms. There was a feeling, by 1985 or so, that the United States had already lost the race, that American industrial champions like Intel, Motorola, and IBM were bound to be left behind by the Japanese microchip juggernaut.

  It was inevitable, in a way, that the U.S. semiconductor community would turn back to its founding father, Noyce, in time of crisis. As the co-inventor of the microchip, he had technical credentials that nobody could gainsay. But he also had a stellar record as a manager, entrepreneur, speaker, and advocate. Almost everything Noyce ever touched had turned to gold; now he was being asked to restore the tarnished luster of American competitiveness.