ARPANET’s First Nodes: The Internet’s Sixties Origin Story

ARPANET did not begin as a public “internet” in the modern sense. In 1969, it was a small, federally funded experiment in connecting research computers across distance. Its first four nodes—at UCLA, the Stanford Research Institute, UC Santa Barbara, and the University of Utah—tested a new way of moving information. From that modest network came technical ideas, habits of collaboration, and protocols that helped make today’s internet possible.

What was ARPANET designed to do?

ARPANET was funded by the U.S. Advanced Research Projects Agency, then called ARPA and now DARPA. Its early goals included secure communication and information sharing among geographically dispersed research facilities. The practical challenge was easy to state but difficult to solve in the 1960s: expensive computers were powerful but isolated, and researchers wanted them to communicate even when they used different hardware and software.

The answer was not simply to put computers on telephone lines. As DARPA’s history of ARPANET explains, the system needed a common way for incompatible machines to exchange data and a design that would not depend on a single node. ARPANET used dedicated telephone lines, but its intellectual leap was to treat the network itself as an active system that could direct small units of information from one place to another.

That purpose also corrects a durable oversimplification. ARPANET was sponsored by a Defense Department research agency, yet its initial nodes were university and research institutions, and its work served a research community. The National Academies’ account of federal networking describes its early use expanding during the 1970s to computer-science researchers and portions of the wider scientific research community. Military funding, academic work, and private contractors all mattered to the story.

Where were ARPANET’s first four nodes?

The first node was installed at UCLA in September 1969. By the end of that year, the original network linked UCLA, Stanford Research Institute (SRI) in Menlo Park, the University of California, Santa Barbara, and the University of Utah. DARPA identifies these four 1969 sites in its timeline; SRI records that UCLA and SRI made the initial proof-of-concept connection and that Santa Barbara and Utah completed the four-node network on December 5.

Each location brought a particular kind of expertise. UCLA’s Network Measurement Center, led by Leonard Kleinrock, was a natural place to study how a network behaved. SRI housed Douglas Engelbart’s Augmentation Research Center and the NLS system, a landmark in interactive computing. According to SRI’s institutional history, its information researcher Elizabeth Feinler later led the ARPANET Network Information Center, which maintained host-name tables, a directory of Requests for Comments, and other tools for making a growing network understandable to its users.

This is a useful surprise in an origin story usually reduced to a few engineers and one failed word: the early internet had an information-organizing problem as well as a transmission problem. A network becomes useful only when people can identify other machines, know what services exist, and share standards. Feinler’s Network Information Center did essential, often less celebrated work of making a technical system navigable.

Why was the first message only “LO”?

At 10:30 p.m. Pacific time on October 29, 1969, UCLA student programmer Charley Kline attempted to log in from a UCLA SDS Sigma 7 host to an SRI SDS 940 host, with Bill Duvall at the receiving end. The intended command was “LOGIN.” The system transmitted the “l” and the “o,” then crashed. UCLA’s Leonard Kleinrock preserves the contemporary IMP log and calls “lo” the first message on the internet; the full login succeeded about an hour later.

The story has survived because it is both comic and revealing. The first famous network exchange was not a polished demonstration or a sentence for the ages. It was an incomplete administrative command. But it showed that two different computers, in separate places, could begin a host-to-host conversation across a network. In technical history, small demonstrations often matter less for their drama than for the new category of possibility they establish.

It is also worth distinguishing dates. The October 29 transmission was an early host-to-host signal between UCLA and SRI; SRI says the link was firmly established on November 21. The four-node network was completed in early December. Treating “the birth of the internet” as a single instant makes a good anniversary headline, but the network’s creation was a sequence of installations, tests, fixes, standards, and new participants.

How did packet switching change computer communication?

ARPANET relied on packet switching. Rather than reserving one uninterrupted circuit for a whole exchange, the method breaks data into small packets bearing destination information. Packets can move through available routes and be reassembled at their destination. The National Academies notes that the technique was proposed as an efficient way to share costly long-distance telecommunications circuits among many users.

The hardware that performed this work was called an Interface Message Processor, or IMP—an ancestor of the dedicated network equipment now called a router. DARPA states that BBN Technologies built the first ARPANET routers under a 1968 ARPA contract. The system’s resilience was important: a circuit-based design could make a failure highly disruptive, while packets could be directed through other available paths. That does not mean ARPANET was magically failure-proof; the “LO” crash is an early reminder that experiments fail. It means the architecture was designed to avoid placing all communication in one fragile route.

Packet switching also helped turn networking into a question of standards. As networks grew, their computers and transmission systems differed. DARPA’s chronology traces later work by Robert Kahn and Vint Cerf on TCP and IP, protocols intended to let separate packet networks communicate. The switch to TCP/IP in 1983 is a key transition: ARPANET was no longer merely a small network, but part of an “internetwork” of networks.

How did ARPANET become the internet people use today?

ARPANET was a foundation, not a finished internet. DARPA’s history charts a 1972 public demonstration, expansion during the 1970s, TCP/IP development, and a 1983 transition as more networks connected. The original ARPANET was formally decommissioned in 1990, while other systems—notably NSFNET for research and education—and commercial providers helped carry connectivity forward. The modern web came later still; it is an application on the internet, not a synonym for every network protocol that preceded it.

Its legacy is visible in everyday activities: remote work, email, file exchange, streaming, messaging, online learning, and the ability of unlike devices to communicate through shared protocols. The history should not become technological destiny, however. The internet’s scale has brought questions its 1969 builders could not settle in advance: privacy, cybersecurity, misinformation, unequal access, platform power, and the energy costs of a permanently connected world.

Those questions give the ARPANET story a present-day edge. DARPA notes that networked devices and mesh approaches can make communication more robust, while connected devices also raise issues of security, privacy, and control over personal data. The old lesson remains relevant: network design is never only engineering. Decisions about who can connect, what information circulates, and who bears risk shape the social life of a technology.

Why It Still Matters

ARPANET’s first nodes show that the internet was made through collaboration rather than a lone act of invention. Researchers, students, programmers, contractors, information managers, and public funders each contributed to a system that had to work across distance and difference. The story of “LO” is memorable, but the fuller history is a lesson in patient collective work.

It also makes the internet feel less inevitable. In 1969, four nodes were a fragile experiment. The network we now take for granted emerged through choices about open protocols, research access, standards, and expansion. Understanding those choices helps citizens recognize that today’s network can be debated, governed, and improved too.

Related reading: Science, Space & Technology pillar page; Silicon Valley’s Sixties Roots: Fairchild Semiconductor and the First Chip.