2007年12月13日 星期四

Serendipity and the Power of Public-Private Partnership

By John Markoff

As a young NASA engineer during the 1980s, Milo Medin liked to irritate his managers by building scientific computer networks using freely available Internet software that outperformed more costly commercial systems.

He was a member of a rebel generation of engineers and scientists that created what would become the commercial Internet during a tumultuous decade. And this group did so by ignoring conventions and adopting a cooperative spirit that turned into the hallmark of the open source software movement.

Some 220 of the original Internet pioneers met in Arlington at the end of last month to celebrate the 20th anniversary of the NSFnet, the scientific data network that was originally constructed to tie together the US' five supercomputer centers and that would ultimately explode into today's Internet. By the time the academic network was shut down in 1996, it connected 6.6 million host computers and extended to 93 countries.

The story of the network and its impact on the world is a case study in the role of serendipity in technology design and in the power of a deftly managed public-private partnership.

Internet pioneers met in Arlington to celebrate the 20th anniversary of NSFnet, the scientific data network that led to today's Internet
"It has to be one of the most successful handoffs in history," said John Marburger III, the science adviser to the president, who spoke at the conference.

At the heart of the partnership's success was a technological gamble: the decision by the National Science Foundation to support a relatively unproven set of software protocols known as TCP/IP that had been developed at Stanford with Pentagon support in the 1980s. It was an idea that is obvious in retrospect, but it was radical at the time.

TCP/IP served as a vital lingua franca between previously incompatible computer networks.

"For the first time in the history of computing, all of the computing platforms spoke the same language," said Allan Weis, a veteran IBM researcher who played a crucial role in the commercialization effort.

Although discussions of industrial policy and the federal government's role in shaping innovation have been muted during the Bush administration, a number of scientists and corporate executives who were at the reunion said NSFnet remained a powerful example of how a handful of government bureaucrats in concert with an equally small number of scientists made a set of carefully considered federal policy decisions, in this case leading directly to the modern Internet.

The researchers who built the original network believe that the NSFnet experience can be translated to advance the country's science and technology policies more broadly. "There are so many important areas where this country could lead," Weis said. "If we learned one thing with the NSFnet experience, I think it was that the government has the ability to help advance science and technology in this country by holding out a carrot and using the stick as a pointer."

The lesson learned, he said, is that it is crucial for government and industry to share financial risks. That would make it possible to move forward in areas as diverse as materials technology, biology and energy efficiency.

The specific form would be to create new "grand challenges" in technology, Weis said.

Though it is almost impossible now to imagine living in a world without the Web, the transition from the academic and scientific NSFnet to the commercial Internet did not come without conflict. There were bitter arguments among the participants over whether commercialization should take place at all.

And when the National Science Foundation contracted with a partnership of IBM, the MCI Corp and the Merit Network -- a group of Michigan universities and a state agency -- to manage the network's backbone, the resulting Non-Profit Advanced Network Services created bitter resentment among early commercial Internet service providers.

"The idea of network as a service was a new thing and it was difficult to convince everybody (a) that it was a good idea and (b) that it was legal," said Steve Wolff, director of network research at the National Science Foundation from 1986 to 1995. A wide range of conference participants said the NSFnet ultimately succeeded because of both the hacker culture of engineers that built the system and the very nature of the network they were creating; it fostered intellectual collaboration in a way not previously possible.

"The model of a network where no one is in charge is a model that can scale," said Douglas Van Houweling, the chairman of the Merit Network when the NSFnet backbone was constructed.

Giving the network time to develop was vital, he said, because the Internet "was an alien concept to the communication industry when it began growing."

While there was no risk for MCI, which was then an upstart trying to gain ground on AT&T, that was not true of IBM. While the company played a crucial role in the development of the Internet, it did so despite the fact that the new network was a direct competitive threat to its multibillion-dollar communications networking business, based on a competing standard known as Systems Network Architecture.

"Although we had the blessing of senior management at IBM, they had no idea how disruptive this would be," said John Armstrong, IBM's director of research at the time the NSFnet was built. "IBM was a large and complex organization, and the decision was made by part of IBM."

In the 2000 election, Al Gore, then the vice president, was derided by opponents who claimed that he had said he "created" the Internet. But many of the scientists, engineers and technology executives who gathered to celebrate the Web's birth say he played a crucial role in its development and expressed bitterness that his vision had been so discredited.

Gore had been instrumental in introducing legislation, beginning in 1988, to finance what he originally called a "national data highway."

"Our corporations are not taking advantage of high-performance computing to enhance their productivity," Gore, then a senator, said in an interview at the time. "With greater access to supercomputers, virtually every business in America could achieve tremendous gains."

Ultimately, in 1991, his bill to create a National Research and Education Network did pass. Funded by the National Science Foundation, it was instrumental in upgrading the speed of the academic and scientific network backbone leading up to the commercialized Internet.

"He is a hero in this field," said Lawrence Landweber, a computer scientist at the University of Wisconsin who in 1980 made the pioneering decision to use the basic TCP/IP Internet protocol for CSNET, an academic network that preceded NSFnet and laid the foundation for "internetworking."

For engineers and scientists like Medin, who went on to be a co-founder of (AT)Home Networks and is now trying to build a national wireless data network, the NSFnet experience provides a lesson about interplay between technology and government policy.

"In that era the government said, `Let's experiment and move everyone forward,"' he said. "If you had waited for a market, it would never have funded an NSFnet."

NY TIMES NEWS SERVICE, VIRGINIA

沒有留言: