Comments on the early days of the Internet from Dr. Stephen Wolff,
based on “Notes from Conversation with Professor David Farber”
As Professor Farber points out, although computer scientists had studied the interconnection of computers, it was CSNET that made such interconnections for purposes other than curiosity-driven research. Contemporaneous was BITNET, an interconnection of academic IBM mainframes that implemented email via punched cards.
Indeed, the NSFNET was originally conceived merely as an adjunct to the NSF’s supercomputer initiative, to interconnect the five NSF supercomputer centers scattered across the US.
But with appetite whetted by CSNET and BITNET, and with the NSF-encouraged growth of the regional networks, traffic on the 56 kb/s NSFNET exploded, leading to a congestive collapse, and the saying that “the NSFNET is so congested nobody uses it anymore”
The second-generation NSFNET, with more than double the original five interconnection points, and 25 times the bandwidth came online in 1988, made possible by a remarkable “public-private partnership” proposed by the University of Michigan, together with the Michigan state government, the academic division of IBM (not the networking division, which remained mired in a hierarchical model of networking), the fledgling telephone company MCI (which had been instrumental in ending the AT&T monopoly), and NSF.
Continued rapid traffic growth on the NSFNET backbone network again led to serious congestion, and in 1992 the bandwidth was expanded by a factor of 30 to 45 mb/s. The success of this expansion was due to a strong public-private collaboration among universities, government agencies, and private companies.
Continued rapid traffic growth on the NSFNET backbone network again led to serious congestion, and in 1992 the bandwidth was expanded by a factor of 30 to 45 mb/s. At the same time, commercial uses of the Internet pioneered in the late 1980s by the first ISPs UUNET and PSINET, and supported by the regional networks, was finally legitimized on the NSFNET backbone by the so-called "Boucher amendment" to the High-Performance Computing Act of 1991.
Throughout the ten-year lifetime of the NSFNET (1985-1995), Professor Farber – who recruited me to come to NSF and work on the NSFNET – was both an enthusiastic supporter and a wise counselor to the infant Internet.
March 12, 2025
Stephen Wolff