The Internet Before It Was ‘The Internet’
In the mid-1980s, the “internet” was a patchwork of disconnected networks, the most famous being the Defense Department's ARPANET. Getting them to talk to each other was a challenge. For scientists, an even bigger problem was access. The nation's most powerful
supercomputers were like isolated islands, accessible only to a privileged few. In 1985, the National Science Foundation (NSF) decided to solve this by funding a new, high-speed “backbone” network to connect its five supercomputer centers and link regional academic networks across the country. They called it the National Science Foundation Network, or NSFNET. Its purpose was simple: give researchers a fast, reliable way to share data and access powerful computing resources, no matter where they were. It wasn't built to be the global commercial network we know today; it was a tool for science and education.
Building a Bigger, Faster Engine
The initial NSFNET backbone, launched in 1986, ran at 56 kilobits per second—about the speed of a dial-up modem. But demand from universities and research labs exploded so quickly that the network was immediately overwhelmed. Recognizing the need for a major upgrade, the NSF partnered with a consortium including Merit Network, IBM, and MCI. By 1988, they had rolled out a new backbone running on T1 lines, a blistering 1.5 megabits per second, or about 25 times faster. This wasn’t just a speed bump; it was a fundamental shift. The network grew at a staggering pace, connecting a few hundred networks in 1988 to over two million computers by 1993. By the early 90s, another upgrade to even faster T3 lines was necessary as traffic soared. NSFNET had become the dominant traffic-carrier and de facto backbone of the American internet.
The ‘No Commercial Traffic’ Problem
There was one big rule on this new superhighway: no for-profit activity. The NSFNET’s Acceptable Use Policy (AUP) strictly limited its use to research and education. You could collaborate on a physics paper, but you couldn't send a purchase order or advertise a product. But the network was so effective that the commercial world grew desperate to get on board. Early commercial Internet Service Providers (ISPs) began to pop up, creating their own networks and workarounds to connect with the larger NSFNET ecosystem. The pressure mounted. By the early 1990s, the NSF began to relax the AUP, acknowledging that the ban was becoming a barrier to growth and innovation. The government realized that private companies could likely provide network services more efficiently and economically than a taxpayer-funded program.
The Master Plan: A Graceful Exit
The NSF’s most crucial contribution wasn’t just building the network; it was planning its own obsolescence. Instead of holding on, the foundation orchestrated a deliberate transition to a privatized, commercial internet. The strategy involved creating a new architecture where multiple commercial backbones would compete and interconnect. To facilitate this, the NSF funded the creation of several key exchange points, called Network Access Points (NAPs). These NAPs were neutral zones where the new commercial backbones could exchange traffic freely, replacing the single, government-run NSFNET backbone. On April 30, 1995, after ensuring the new commercial system was stable, the NSF officially decommissioned the NSFNET backbone. The government-run training wheels came off, and the internet was handed over to the private sector.
