A Network for Scholars, Not Shoppers
In the mid-1980s, the National Science Foundation created NSFNET to connect researchers at universities with a handful of powerful supercomputer centers. It was a revolutionary upgrade from its predecessor, ARPANET, and quickly became the backbone of the nascent
internet in the United States. But it came with a crucial catch: an "Acceptable Use Policy" (AUP). This policy strictly limited the network's use to research and education. You could use it to debate astrophysics, but not to sell books or manage a payroll. For-profit activity was explicitly banned. For nearly a decade, the very heart of the internet was a commerce-free zone by design.
The Commercial Pressure Cooker
By the early 1990s, the digital world was straining against its academic confines. Businesses saw the immense potential of this interconnected system, and traffic on NSFNET was exploding. The NSF realized that running the internet's backbone wasn't a sustainable long-term government project. Commercialization was inevitable, but how it would happen was the million-dollar question. The partners who managed the NSFNET backbone—Merit Network, IBM, and MCI—proposed a solution that set the stage for a major conflict. They formed a new nonprofit called Advanced Network and Services (ANS). This new entity would manage the upgraded backbone, but it also created a for-profit subsidiary, ANS CO+RE, designed to sell network access to commercial customers.
The Fork in the Road: A Single Tollbooth?
The ANS proposal was controversial because it threatened to create a de facto monopoly. Here was a group, born from the government-funded project, that now owned the core infrastructure and was also competing as a commercial provider. Other fledgling internet service providers (ISPs) were terrified. They argued that ANS would have an unfair advantage, potentially controlling access and pricing for the entire internet. They envisioned a future where one company operated the main highway and also owned the only gas stations on it. This was the moment things could have gone a completely different way. Instead of an open market, the internet could have been dominated by a single, private gatekeeper, fundamentally altering the competitive landscape before it even began.
The Fight for an Open Marketplace
The backlash from the wider networking community was swift. Competing commercial providers formed groups like the Commercial Internet eXchange (CIX) to create alternative pathways for traffic, deliberately routing around the ANS-controlled backbone. They lobbied for a different transition model—one that would dismantle the government-funded backbone entirely and replace it with a system of competing, private backbones. The key to making this work was the creation of neutral connection points. In this vision, the government wouldn't hand the keys to a single successor; it would create public squares where any and all commercial networks could meet to exchange traffic on equal footing. This was a battle between a vertically integrated, closed model and a decentralized, open-market one.
How the Open Internet We Know Was Born
Ultimately, the open-market vision won. The U.S. government, through the NSF, decided against privatizing the backbone into a single entity. Instead, in 1995, it officially decommissioned the NSFNET backbone. In its place, it funded the creation of several Network Access Points (NAPs), the neutral meeting places the CIX and others had advocated for. Companies like Sprint and Ameritech were tapped to run these NAPs, creating the foundation for a competitive marketplace. Private companies would now build and manage their own long-haul networks, and the NAPs would ensure they could all talk to each other. On April 30, 1995, the Acceptable Use Policy officially died, and commercial traffic was free to flow across the entire internet.
