Ignore the Supercomputer Race
A new list of the world's 500 fastest supercomputers suggests that China might be speeding past the U.S. in the race for technological supremacy. China now holds the two top spots, and placed a total of 167 machines on the list. The U.S. had only 165 on the list, with its fastest placing a very distant third.
That's leading some American commentators to wring their hands. Wired went so far as to declare a "blow out" in the race for supercomputer supremacy. But as impressive as China's accomplishment is, there's no reason to panic. The race for technological dominance won't be won by measuring who can build faster computers. Instead, what will matter is who invests most wisely in basic research -- the kind of methodical, unglamorous science that might only yield results years in the future.
The immediate goal of such research isn't necessarily a product. But long-term, it might turn into many. Government-funded work on three-dimensional seismic imaging, for instance, helped lay the groundwork for the fracking revolution of recent years. The Human Genome Project, started in 1990, will provide scientists with raw material to cure diseases for decades to come.
"People cannot foresee the future well enough to predict what's going to develop from basic research," is how George Smoot, a Nobel Prize winner in physics, once explained it. "If we only did applied research, we would still be making better spears."
In this, the history of supercomputers is instructive. Bell Labs was doing fundamental research on semiconductors all the way back in the 1940s. Eventually, what they developed was licensed to other companies, including Texas Instruments, which then developed transistors, integrated circuits and other components. It wasn't until the early 1960s that such technology coalesced into an early version of the supercomputer.
The U.S. dominated supercomputing for two decades, but it was only a matter of time before others piggy-backed on established technology to catch up. In 1981, Japan started a government-backed initiative to develop its own machines. China did the same (with World Bank backing) in 1989. Russia, the EU and several European countries have joined the game as well.
Amid such competition, the title of world's fastest supercomputer tends to be a fleeting honorific. And the wisdom of engaging in the race has always been questionable. In 2010, President Barack Obama's council of science and technology advisors argued that a "single-minded focus" on increasing speed diverts resources from more creative approaches to computing. In most fields of science and engineering, for instance, performance improvements from more sophisticated algorithms -- the mathematical rules used to solve a problem -- have topped those from faster processors in recent years.
That kind of inventiveness is often the result of years of patient (and unprofitable) research -- a lesson America shouldn't forget. Although the U.S. still leads the world in R&D funding, as a percentage of gross domestic product its efforts now lag behind South Korea, Japan and Taiwan. The portion of the federal budget dedicated to research and development has been in decline since 1965. Corporate basic research has likewise been falling, losing out to the product-oriented kind demanded by shareholders and the global marketplace.
Meanwhile, the competition for basic research is heating up. Chinese research programs have historically focused on hitting clearly defined goals, which is one reason that supercomputer speed has been such an appealing benchmark: In 2012, 84 percent of China's R&D went to the commercialization of technologies. But policymakers are starting to change their tune. In mid-June, China's National Science Foundation announced big funding increases for basic research, including cosmic ray physics, mathematics, brain science and infectious diseases.
This is no bad thing. Invigorated competition for basic research would be far more productive than the race to soup up supercomputers. Fields such as synthetic biology, quantum computing and photonics all stand to benefit from a healthy international rivalry. And areas with less obvious payoffs may prove even more important. Current mysteries such as dark matter might one day turn out to be just as fruitful as one-time mysteries like radio waves. The benefits may not materialize for years or even decades. But if history is any guide, it's a good bet that they'll be worth the wait.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Adam Minter at firstname.lastname@example.org
To contact the editor responsible for this story:
Timothy Lavin at email@example.com