Computer enthusiasts often describe the Internet as an organic creation, more like electronic kudzu than manmade artifacts like television or soda pop.
Computer networks generally and the World Wide Web in particular owe their origins to human ingenuity and enterprise, of course, but no single individual is credited with the invention: There is no Thomas Edison of the Internet.
Many people over the years have hooked computers together one way or another to enhance their usefulness without actually setting out to invent the Internet. Even now, no one knows with certainty just how the Web’s evolution will continue, although many try their best to shape it to their advantage.
But having said that, a strong case can be made that work done on the Urbana campus of the University of Illinois over the past decade has been the major force shaping today’s Internet. Without such innovations as Telnet and Mosaic, developed at the U. of I.’s National Center for Supercomputing Applications, there would be no Web.
And even though his name isn’t as widely known as those of computer luminaries such as Bill Gates and Larry Ellison, the founder of NCSA, Larry Smarr, may well be the one person whose vision has had more to do with creating computing as we know it than any other.
Oddly, while Smarr’s vision helped make the Web possible, he never dreamed of making millions on the Internet. He still doesn’t, even as he begins a new undertaking aimed at making virtual reality machines as common a decade from now as Web-connected computers are today.
Smarr is a scientist so smitten with the overwhelming potential of a technology that spreading his vision has become almost a religious mission.
Indeed, he can even pinpoint the revelation that changed his life, which he likens to “what happened to Paul on the road to Damascus” when the one-time persecutor of Christians became their leading advocate.
A native of Columbia, Mo., Smarr, who is now 48, followed his father’s advice as he started taking classes at the University of Missouri and began learning about computers as well as physics and math.
Smarr soon became fascinated with Einstein’s theory of general relativity and the possibility of using computers to work through the complex equations that could explain such esoteric phenomena as the behavior of black holes in space.
Smarr went on to doctoral work at the University of Texas, where his doctoral adviser, Bryce DeWitt, suggested he would have to get a top security clearance from the military to pursue his interests. This came as a shock to the long-haired, decidedly unmilitary Smarr.
“The problem was there was no commercial market for supercomputers in the 1970s,” said Smarr. “The military needed them for code encryption and designing bombs, and would spend any amount to get them. So if you wanted to use the computers, you had to have clearance.”
To many people working in the nation’s weapons labs, bomb building was their day job and general relativity was their hobby, Smarr said.
“I never became a weapons designer, but I hung with them doing relativity,” he said.
Smarr’s revelation came shortly after he got his top secret clearance and was admitted to the Lawrence Livermore National Laboratory in California, where the scientists showed him a computer simulation of a hydrogen bomb detonation.
Seeing that it was possible for a computer to work out complex equations describing natural phenomena and represent the results in moving images was like a blinding flash of light to Smarr’s intellect.
“Anything in nature could be solved and visualized using computer graphics,” he said. “And all branches of science could make great leaps. Yet access to this technology was limited to the inner sanctum within the fortress walls. No one on the outside even knew it was possible.
“It was as if in the 17th Century Isaac Newton had invented calculus and then told the British Navy this was a great edge for navigation so they kept it a secret from the rest of the world.”
During a post-doctoral odyssey of Ivy League schools, he spent summers at Livermore, where he put in 100-hour weeks running equations on supercomputers. Smarr tried to convince colleagues at Harvard University there was a great new tool out there that beat the heck out of pencils and paper.
“They looked at me like I was from Mars,” said Smarr. “These were top scientists, Nobel Prize winners, and they had no concept of supercomputers.”
When he joined the University of Illinois in 1979, Smarr did find some soulmates who hungered for more computing power. As home to Illiac, an early room-sized computer made with vacuum tubes, Illinois had a tradition in computing shared by few other universities.
Smarr and some other faculty members pooled their departmental resources to build a homemade version of a supercomputer using available commercial mainframes. He wrote a program to model black holes that took their new computer eight hours to run.
Later, when Smarr had the chance to take that same program to Germany to run it on a Cray supercomputer at the Max Planck Institute in Munich, it took two minutes to run. One German academic teased Smarr about having to come from the rich United States to borrow time on a supercomputer made in the U.S. and asked him how such a country could have won the war.
That reignited Smarr’s desire to unleash America’s supercomputing power for the nation’s scientists and engineers. He started calling scientists around campus and asked them what specific things they might accomplish if only they had enough computing power.
“We documented 65 scientists on our campus whose work was blocked by lack of computing power,” Smarr said. “I wrote the proposal and didn’t know where to send it because there was no department within the National Science Foundation to handle it, so I just sent it to the director.”
A similar lament also was received at NSF from a scientist in San Diego. Several reports decrying America’s lagging basic research were coming out at the time. In response, the NSF decided to finance five supercomputer centers around the country. Smarr became director of the one in Urbana that opened in 1985.
From the first, Smarr believed that you shouldn’t have to be a computer wizard to use computer technology and that you should be able to work from your desktop, wherever your office was, rather than come to a central place where the computer was housed.
“Our vision was that the Cray was the big supercomputer hiding behind the Mac sitting on your desk,” said Smarr.
Another concept Smarr pushed from the beginning was graphical interfaces and visual presentation of information. The center brought people from the university’s art department onto its team to help plan how to best present information visually so that people could spot subtle trends and interrelationships buried in the data.
As soon as he got the center running, Smarr was traveling to Chicago and other commercial centers to seek industrial partners who would contribute to the center and in return receive help in applications useful to their business.
Partners included Eli Lilly & Co., the pharmaceutical firm based in Indianapolis. Lilly’s subsequent use of computer modeling to design new drugs has helped pushed the entire industry into a new era.
“NCSA has been a rare success in America of a partnership among the federal government, the universities and private enterprise,” said Riaz Abdulla, senior information adviser to Eli Lilly. “It has achieved the major goal of spreading high performance computing to American business. Larry Smarr is a total visionary.”
Other partners have been Motorola Inc. of Schaumburg, Caterpillar Inc. of Peoria and Sears, Roebuck & Co. of Hoffman Estates.
At Sears, the NCSA partnership has proven very valuable, said Chris Herr, a Sears technical consultant.
“Outside computer companies worked for two years on projects to automate manually intensive processes in the stores and just couldn’t get anything satisfactory,” said Herr. “In working with NCSA, we came up with working proof of concept for a system in just six months. It’s been quite a valuable asset to us.”
In the early days of NCSA, one of the most dramatic uses of supercomputer power was explored by Bob Wilhelmson, a meteorologist on the Urbana campus who modeled thunderstorms and tornadoes.
“Those first models of thunderstorms demonstrated to many people what seeing the hydrogen bomb detonation did for me,” said Smarr. “It showed that we could do science on the computer.”
It also helped to give tens of thousands of scientists and engineers an insatiable appetite for more computing power. Science has always engaged in oversimplification of reality in order to understand tiny pieces of nature. But whereas engineers five years ago modeled turbulence in static two dimensions, they now can do it dynamically in three dimensions.
“This will soon filter into industry,” said Danesh Tafti, an engineer at NCSA. “We are still a long ways from modeling full-fledged reality, but we are getting much closer.”
Earlier this year the NCSA embarked on a new existence. The center in Urbana and another in San Diego have been designated by NSF to promote the next phase of supercomputer use.
This includes initial funding in the range of nearly $30 million from the federal government, which is almost double what the center received in the past. It also gives the center the official mandate of providing computer learning to high schools, something it did in the past as a sideline.
Using virtual reality technology developed at the University of Illinois at Chicago campus by a team led by Tom DeFanti, Smarr has plans for a new university-based network of advanced graphics computing, called the Grid.
Whether this will lead to a commercial success comparable to the Web, Smarr said he doesn’t know or care. His interest is in bringing the power of computing to scientific research, industrial development and classroom learning.
“We build things that render nature in visually exciting ways,” he said. “When people see it, they say `I want my stuff to look like that,’ and it spreads that way.”
INTERNET OVERCAME OPPOSITION FROM BIG PHONE FIRMS EARLY ON
Early efforts by academic scientists to link computers across the country in a single network financed with federal money were stymied by the big telecommunications companies.
“The suits from AT&T and MCI said it was inappropriate for the government to build something like that,” said Larry Smarr, director of the National Center for Supercomputing Applications at the University of Illinois.
“The problem was, there was no commercial market for such a thing, so the telephone companies weren’t building it either,” Smarr said. “It was a stalemate for us because the National Science Foundation doesn’t have enough clout to take on the phone companies in a political battle.”
A forerunner to the Internet, Arpanet, was already in place, but required users to have a high level of skill in UNIX computing and was used mostly by military researchers and computer scientists. Smarr’s proposal was for a network that could be used by researchers who weren’t UNIX wizards.
Smarr and the directors of the country’s other four supercomputer centers met in Colorado at the National Center for Atmospheric Research to discuss the problem in 1986. He proposed that only the five supercomputing centers and NCAR be linked in a supercomputer network.
“The telco guys asked us how many people might use such a network, and we said probably about 100,” said Smarr. “They figured that was no market they cared about, so they said OK, that’s fine. But once we got linked, some regional centers asked to hook up and pretty soon we reached a critical mass.”
As the big telephone companies saw they could make money from the computer linkups, they joined in supporting it. Once network use spread from academics and computer hobbyists to businesses and the general public, NSF funding was practically nil. The government is now out of the picture.
“It’s the perfect model for how government money should seed something too new for private funding and then back off,” said Smarr.
