Can Putin Survive?
The Lessons of the Soviet Collapse
In June, the Sunway TaihuLight, a Chinese supercomputer, rose to the top of an international ranking of the world’s high-performance computers. Chinese computers have led TOP500's rankings for years, but the TaihuLight is different: unlike most of China’s other superfast computers, which use processors designed in the United States, it was built with the Chinese-made SW26010 many-core processor, supplied by the Shanghai High Performance IC Design Center. In other words, the new rankings showed that China’s domestic computing industry has come into its own.
The news of this homegrown success may come as a surprise to those who question the ability of China’s indigenous high-tech sector to innovate. Yet China’s computing industry has a long history of going it alone.
In fact, 44 years ago this summer, a group of U.S. citizens visited China in what was likely the first delegation of American computer scientists to make a trip to the People’s Republic. Arriving in China only months after the epochal February 1972 visit to the country by U.S. President Richard Nixon, the team came to learn more about what China’s computer engineers had been up to during the many years of U.S.–Chinese estrangement.
In 1958, a group of engineers at the Institute of Military Engineering in Harbin had created China's first vacuum-tube computer. The country's computing industry advanced in fits and starts after that, initially with the support of Soviet engineers. Then, in 1960, the Sino-Soviet Split dealt Chinese computing a major blow, as Moscow rapidly withdrew its advisers—and their technical expertise—from China. Yet in the decade that followed, Chinese engineers managed to carry on alone. By the time the American delegation arrived in Guangzhou in July 1972, China had developed a computing industry capable of producing a third-generation computer—a smaller, faster breed of machine based on integrated circuits rather than individual transistors. What was even more surprising: these integrated circuits were being built domestically.
COMPUTING WITH CHINESE CHARACTERISTICS
The American delegation was the brainchild of a young computer scientist named Severo Ornstein, then a systems design engineer at Bolt, Beranek, and Newman, a high-tech firm in Cambridge, Massachusetts. Ornstein was an avid traveler, he told me in a recent conversation, and China was high on his list of destinations—but as a relatively unknown junior engineer, he knew that Beijing was unlikely to issue him the historic invitation such a trip would require.
So Ornstein used an amusing trick to improve his odds. He composed a list of the most celebrated computer scientists in the United States, he explained, and called them up one by one. When the scientists asked who else would go on the trip to China, he read off the names of everyone on his list—regardless of whether they had signed up at that point. The ruse worked: a number of prominent computer scientists agreed to join the delegation, resulting in a thick packet of powerhouse CVs that Ornstein and two of his colleagues presented to the Chinese embassy in Ottawa, Canada, in 1971.
The Sino-Soviet Split dealt Chinese computing a major blow.
Ornstein waited for months without hearing from the Chinese. As the academic year wound down and potential members of the delegation prepared for summer travel and research, it seemed that the project was about to fizzle. But in April 1972, a letter arrived at Ornstein’s Boston office, bearing good news: Ornstein, five colleagues, and their wives had been invited to come to China in July. The visit would last three weeks and would be paid for in full by the Chinese government.
The delegation that Ornstein assembled was a blue-ribbon panel of U.S. computer science. Along with Ornstein, the team included Herbert Simon, a Nobel laureate and professor of computer science and psychology at Carnegie Mellon University; Alan Perlis, a professor of computer science at Yale University and the inaugural recipient of the Turing Award; Anatol Holt, the director of the Information Systems Theory Project of the Massachusetts Computer Associates, a subsidiary of the pioneering software company Applied Data Research; Wesley Clark, the former director of Washington University in St. Louis’ Computer Systems Laboratory and a widely admired developer of advanced computer systems; and Thomas Cheatham, then the director of the Center for Research in Computing Technology at Harvard University.
On July 10, the delegation arrived in Guangzhou. Over the course of the next three weeks, the scientists visited the Shanghai Computing Research Institute and the Institute of Computing Technology in Beijing, two of the main centers of computer science in China at the time.
Despite China’s relative scientific isolation in the years after the Sino-Soviet split, the Americans found that Chinese engineers had come a long way. The subtitle of the team’s report, which they published in Science a few months after their return to the United States, captured this impression: “Computer technology advances rapidly in China with no external aid,” it read. The Americans’ Chinese hosts made a point of emphasizing what they termed “the policy of self-reliance and keeping development in our own hands,” as Mary Allen Wilkes, a pioneer in personal computing, recalled in her diary. (Wilkes accompanied the delegation to China through her marriage to Wesley Clark but was a prominent computer scientist in her own right.) Self-reliance was less a matter of national pride than it was a necessity, the Chinese engineers told the Americans. “We have had to work by ourselves to develop the transistorized and integrated circuit computers,” one member of the Institute of Computing Technology in Beijing explained, “and we have brought up a generation of scientists.”
As for the Chinese computer scientists’ technical achievements, they were impressive. In 1965, the Beijing Institute of Computing Technology completed work on the model 109C computer, which boasted a performance of 115 kiloflops. That speed paled in comparison with Control Data Corporation’s 6600 computer, a U.S.–based machine that at three megaflops was then the world’s fastest computer—yet it still testified to the strides that China had made without technical support from either of the Cold War superpowers. Three years later, in the midst of the Cultural Revolution, Chinese factories began to manufacture integrated circuits—a key ingredient in third-generation computing. By 1970, the institute in Beijing had produced its model 111 computer, a system reaching performance speeds of 180 kiloflops. The same year, the Shanghai Computing Research Institute repurposed a local factory once used to produce window components for the production of an integrated circuit digital computer—a further step in the country’s development of a computing industry. Still referred to as the “window handle factory” despite its new purpose, the factory continued to employ many of the same women who had worked there before.
The Americans’ Chinese hosts were explicit about their interest in a particular kind of computing. As the delegation’s report explained, “The Chinese we talked to indicated a strong interest in what they called the ‘super computer.’” (The inclusion of scare quotes around the term “super computer” is revealing: it suggests that the expression, coined years before by American computer scientists, was novel enough at the time to merit qualification.) Chinese computer scientists, the report noted, inquired about “very big and very fast machines such as the CDC STAR computer and the Burrough’s B6700.”
The Americans were notably surprised by their hosts’ emphasis on supercomputing, which the Chinese made clear they were employing for a number of typical purposes, such as weather forecasting and computational mathematics. (The delicacy of U.S.–Chinese relations meant that supercomputing's other classical applications—cryptography and the design of nuclear weapons—were not discussed during the trip.) “By way of contrast,” the report continued, “they showed little interest in mini-computers, which have become prevalent in the United States in recent years and which, because of their simplicity and economy, have made possible many new applications.” The report then made a striking prediction: “One guesses they will continue the trend towards bigger and faster computers… perhaps attempting a very large step next.”
The news surrounding the Sunway TaihuLight is evidence of the report’s prescience. And the speech that Guo Moruo, the president of the Chinese Academy of Sciences, delivered to the American team as they prepared to leave China offers another reminder for the present. “In introducing his toast to friendship,” Wilkes recalled in her diary, “[Guo] summarize[d] what we have seen and speaks of China’s technical backwardness, saying that while China’s computers perform a hundred thousand calculations per second, the United States computers operate at a hundred million per second. China’s technology, he said, has a great deal to learn from the United States.” Nearly a half century later, it seems that the United States still has a great deal to learn from China.