The DNS Fight That Nearly Broke the Internet Before It Even Started

The DNS Fight That Nearly Broke the Internet Before It Even Started

In 1987, the internet was small enough that the people who ran it could fit in a single conference room. And they spent most of a weekend screaming at each other about email addresses.

That argument is the forgotten history of how the internet almost didn't happen—because nobody could agree what it was for.

The Two Tribes

There were really two camps in the early days, and they barely spoke the same language.

The ARPANET crowd had been building a military-academic network since 1969. They cared about reliability, hierarchical control, and keeping things stable. To them, the internet was a tool for research—a fancy way to share computing resources between universities and defense contractors. If you weren't a scientist or a soldier, you didn't belong.

The Usenet/Unix people were the counterculture. They grew out of the late-70s hobbyist scene where anyone with a Unix box and a phone line could dial into bulletin boards. They believed the internet was a conversation. Anyone should be able to run a server, host a topic, and talk to strangers. The phrase "information wants to be free" came from this side.

The Great Address Schism

The fight that nearly tanked everything was over domain names.

In the early 1980s, the ARPANET used a single file called HOSTS.TXT—a flat list of every computer on the network, maintained by one guy at SRI. If you added a machine, you emailed him, and he updated the file. This worked when there were 200 machines. By 1983, there were 500. By 1986, over 2,000.

The ARPANET side proposed a centralized system: one root server, one authority, you get a name if they approve your use case. If you were running a research lab, fine. If you were some guy with a Commodore and a modem, forget it.

The Usenet side wanted a distributed system. Let anyone register a domain. Let communities self-govern. The only rule: don't break the network.

The argument got nasty. People accused each other of trying to "own the internet." There were threats to fork the entire network—create a separate DNS for hobbyists and one for "official" users.

What Actually Broke the Deadlock

The solution came from an unlikely source: the Internet Engineering Task Force (IETF), which was itself barely a formal organization. They decided to hold a vote. But they didn't count hands—they counted email messages.

Over three months in 1987, participants posted their arguments to a mailing list. The ARPANet side wrote detailed white papers about reliability and governance. The Usenet side wrote short, sarcastic rants about "network fascism."

What finally tipped the balance was a demonstration. One Usenet programmer—a grad student named Paul Vixie—set up a small DNS server on a borrowed Sun workstation and let anyone register a domain. In six weeks, he had more domains than the official ARPANET root had in three years.

The ARPANet leadership realized they'd already lost. If they didn't adopt a distributed model, a parallel internet would simply exist alongside theirs—and they'd be the ones left out.

The Compromise That Built the Web

The IETF produced RFC 1034 and 1035 in November 1987. It was a careful compromise:

• Centralized root servers (ARPANET got its authority)
• Decentralized domain registration (Usenet got its freedom)
• No gatekeeping on use case (Anyone could register a .com, .org, or .net—including that guy with the Commodore)

That last point is why you're reading this article on a domain that cost $12. If the ARPANet crowd had won, PythonSkillset.com would have needed approval from a committee of university professors before it could exist.

The Ghosts of That Argument Today

The fight never really ended. It just went underground. Every major internet controversy since then—net neutrality, walled gardens, the battle between open platforms and closed apps—is a rehash of that 1987 argument.

The man who maintained HOSTS.TXT, a programmer named Elizabeth "Jake" Feinler, later said: "We thought we were building a tool for scientists. Then people started using it for recipes. Then for flirting. Then for complaining about their bosses. We didn't plan any of that."

The early pioneers argued because they genuinely believed the internet would be ruined if the other side won. One side was right. The other side was also right. And the messy, screaming, deeply-human argument they had in that conference room is why the internet is both reliable enough for banking and chaotic enough for cat memes.

If you ever feel frustrated by internet governance, remember: the very existence of governance is a miracle. We came this close to having two internets, one for "important" people and one for everyone else. We chose the one for everyone.