mostly interested in theoretical physics, astronomy, and cosmology. He got his Ph.D. in astrophysics from the University of Texas at Austin, developing mathematical techniques needed to conduct computational-based research—using supercomputers to analyze the dynamics of such phenomena as colliding black holes. But he notes ruefully, “When I got my Ph.D. in ’75, my Ph.D. advisor said, ‘OK, now you need to get a top secret clearance to do nuclear weapons research.’ I said, ‘But my research is in open science.’ And he said, ‘The only place where we have supercomputers are the national nuclear weapon design laboratories like Livermore and Los Alamos.’”
Instead, Smarr says he left in the late 1970s for Germany’s Max Planck Institute for Physics, where he had the opportunity to conduct “open” research in relativistic astrophysics, using the first Cray supercomputer in Europe.
After he returned to the U.S. in the early 1980s, Smarr says he wrote an unsolicited proposal to the National Science Foundation for a $55 million grant to establish a national supercomputer center. He planned to conduct open scientific research and wanted to make a U.S. supercomputer available to scientists and universities throughout the country. The NSF decided instead to hold a nationwide competition to establish five supercomputer centers, and Smarr learned his proposal would be considered as part of that competition, along with a similar proposal that had been submitted for a supercomputer center in San Diego. The NSF funded both proposals in 1985, which is how the San Diego Supercomputer Center was established at UCSD at the same time as the NCSA in Illinois.
At about the same time, Smarr says, “we got into saying, OK, well, don’t we need to connect these supercomputers?”
For Smarr, this became one of the key events in the development of the Internet. In the mid-80s, he says most computer application scientists, especially the physicists with clout at the NSF, used the instruction set architecture developed by Digital Equipment Corp., which was known as the DEC VAX. “The natural thing would have been to connect [the supercomputers] with the commercial, proprietary DEC-net,” Smarr says.
But Smarr says he argued “very strongly” for the NSF to instead adopt an open standard for its set of network communications protocols, which was known as Transmission Control Protocol and the Internet Protocol, or TCP/IP. “I was only one of the people that was arguing for that,” Smarr says. “But I was pretty persistent.”
In fact, Smarr says, “We convinced the NSF to make a ruling that