Well-VersedLibraryCoverageAbout
2026-06-23 · Technology
Technology · The Birth of the Internet

ARPANET

The network that became the internet went live in 1969 with a message meant to say login that managed only its first two letters.

Front panel of an Interface Message Processor, with a brass plate reading 'Interface Message Processor, developed for the Advanced Research Projects Agency by Bolt Beranek and Newman Inc.'
An Interface Message Processor, the machine built by Bolt Beranek and Newman that served as each ARPANET node's gateway. One stood at every site, shuttling packets so the host computers did not have to. Source: Wikimedia Commons, where the author and licence terms for this file are recorded.

At half past ten on the night of October 29, 1969, a UCLA graduate student named Charley Kline sat at a terminal and tried to log in to a computer hundreds of miles up the coast at the Stanford Research Institute. He typed an L, and over an open phone line a programmer named Bill Duvall confirmed it had arrived. He typed an O. That arrived too. He typed a G, and the distant machine crashed. The first message ever sent between two computers on the network that would become the internet was therefore a single accidental syllable, lo, as in lo and behold.

ReadUCLA IMP log, 29 October 1969

22:30   Talked to SRI
Host to Host   CSK

Charley Kline's entry in the UCLA Interface Message Processor log, written the night of the first ARPANET transmission.

The network was called ARPANET, after the Advanced Research Projects Agency that paid for it, a Pentagon office created in the alarm after Sputnik to keep American science ahead of the Soviets. Its purpose was practical and expensive. Computers in 1969 were rare, enormous, and mutually incompatible, and a researcher who wanted time on a machine in another city often had to fly there. J.C.R. Licklider, the psychologist turned administrator who shaped ARPA's computing program, had been sketching since 1962 what he half-jokingly called an Intergalactic Computer Network, a way for distant machines and the people at them to share work as though they sat in one room. Lawrence Roberts turned that vision into a wiring diagram and a budget.

The hard problem was how to move data reliably between machines over unreliable phone lines, and the answer reordered the century. Rather than open one continuous circuit between two computers, the way a telephone call ties up a line from end to end, ARPANET chopped every message into small standardized chunks called packets, stamped each with its destination, and let them find their own way across the network before being reassembled at the far end. Paul Baran at the RAND Corporation and Donald Davies at Britain's National Physical Laboratory had reached the idea separately in the early 1960s; Davies supplied the word packet, and Leonard Kleinrock at UCLA built the mathematics that showed it would hold up under load. A network put together this way has no single throat to cut: when one path fails, the packets simply route around it.

Making it real fell to Bolt Beranek and Newman, a Cambridge consulting firm whose engineers spent 1969 turning a refrigerator-sized Honeywell minicomputer into the device pictured above, the Interface Message Processor. One IMP sat at each site as the network's gatekeeper, managing the packet traffic so the host computers were free to do their own work. The first went to Kleinrock's lab at UCLA that September. By the end of the year four were running, linking UCLA, the Stanford Research Institute, the University of California at Santa Barbara, and the University of Utah.

From there it spread quickly. Fifteen sites were connected by 1971, and within a few more years the map had become a dense web reaching coast to coast and across the Pacific to Hawaii. The application that made the network feel indispensable was one nobody had ordered up. In 1971 a BBN engineer named Ray Tomlinson wrote a program to pass messages between users on different machines and reached for a spare key to separate the person's name from the computer's, handing the world the @ sign in every email address. Inside a couple of years, electronic mail was carrying most of ARPANET's traffic.

Hand-drawn logical map of the ARPANET in March 1977, showing dozens of host computers connected across the United States and to Hawaii.
The ARPANET in March 1977, drawn as a logical map of its hosts. The four-node network of 1969 had become a continental system; each circle marks an Interface Message Processor, with the host machines hung off it. Source: Wikimedia Commons, where the author and licence terms for this file are recorded.
· · ·

ARPANET was a single network, and the internet is a network of networks; the bridge between the two ideas was a pair of protocols. Across the 1970s Vint Cerf and Bob Kahn worked out TCP/IP, a common language that let networks built by different people on different hardware interconnect as equals. On January 1, 1983, ARPANET switched over to it in one coordinated cutover its operators called flag day, the moment usually marked as the technical birth of the internet. The original network was quietly retired in 1990, by then a small backwater of the thing it had set in motion. Its founding bet still governs the network you are reading this on: keep the middle of the network simple and push the intelligence out to the edges, into the machines themselves, so that nobody at the center has to approve what passes through.

Up next
The Handshake
From a ninth-century-BC Assyrian relief to the modern grip: where the gesture came from and how to do it well.
Read →

Images on this page are reused from Wikimedia Commons. Each is linked to its file page, where the author and the licence terms for that file are recorded. Thumbnails in “Up next” are credited on the lesson they link to.