From ARPANET to the Internet: How One Protocol Ate Every Network
ARPANET was one research network among several incompatible experiments in the 1970s. TCP/IP is the specific technical decision that let it absorb all the others into a single, unified internet.
Before there was an “internet,” there were several separate, mutually incompatible research networks — ARPANET among them, but not alone. What actually produced today’s single, unified internet wasn’t any one network’s victory over the others; it was a protocol designed specifically to connect fundamentally different networks together.
ARPANET’s actual starting point
ARPANET, funded by the US Department of Defense’s Advanced Research Projects Agency, went live in 1969, connecting a small number of university and research computers using packet-switching — breaking data into small, independently routed packets rather than requiring a single dedicated circuit, the same core idea underlying networking today.
The problem TCP/IP was specifically built to solve
By the mid-1970s, ARPANET was just one of several incompatible packet-switched networks in existence — different research groups and countries had built networks using entirely different underlying technical approaches, with no shared method for one network to talk to another at all.
TCP/IP, developed principally by Vint Cerf and Bob Kahn and finalized through the late 1970s, was explicitly designed as an “internetworking” protocol — one that didn’t require every network to use identical underlying technology, only that each network implement a common protocol at its edges for passing data between networks.
January 1, 1983: the actual cutover
ARPANET formally switched from its earlier NCP protocol to TCP/IP on January 1, 1983 — an event sometimes called “flag day” among the small community of people running the network at the time, since every connected system had to switch simultaneously for the network to keep functioning at all.
Why this specific protocol, not a specific network, is the real ancestor of the internet
Once TCP/IP existed as a common language, entirely separate networks — university networks, corporate networks, and eventually consumer internet service providers’ networks — could all interconnect by implementing the same protocol, without needing to share ARPANET’s specific underlying infrastructure or governance. The “internet” that resulted is, definitionally, a network of networks unified by this shared protocol, not an expansion of ARPANET itself into something larger.
What happened to ARPANET itself
ARPANET, the original specific network, was formally decommissioned in 1990 — by which point it represented a small and no longer especially significant fraction of the much larger TCP/IP-based internet that had grown up around and beyond it, using the same protocol but largely different physical infrastructure.
The lesson this history actually supports
The internet’s actual origin story is less “one research network grew into everything” and more “a protocol designed for interoperability between different networks let a great many separately built networks merge into one” — a distinction that matters because it’s the protocol’s design philosophy (interoperate rather than replace) that scaled, not any particular network’s own infrastructure or institutional lineage.