The first message sent over the internet was an accidental "LO"
In October 1969, researchers at UCLA tried to send the first message to Stanford over ARPANET, the precursor to the internet. They planned to type "LOGIN." However, the system crashed after they typed the first two letters. About an hour later, after fixing the bug, they successfully transmitted the full word.
The October Night in Boelter Hall
On the night of October 29, 1969, a small team of computer scientists gathered in Boelter Hall at the University of California, Los Angeles. Their goal was straightforward but technically unprecedented: establish the first remote host-to-host connection between two separate computer systems on the newly constructed ARPANET. At UCLA, student programmer Charley Kline sat at an SDS Sigma 7 computer terminal under the direction of Professor Leonard Kleinrock. More than three hundred miles to the north, at the Stanford Research Institute in Menlo Park, programmer Bill Duvall sat before an SDS 940 mainframe.
Because neither side had any visual or automated indication of what the remote machine was receiving, Kline and Duvall maintained an active telephone voice link throughout the test. They wore telephone headsets, speaking back and forth to confirm that each keystroke made in Los Angeles was properly registering in Northern California. The plan was to log in remotely to the SRI machine by typing the command word LOGIN.
The Crash at the Third Keystroke
The transmission began at approximately 10:30 PM. Kline typed the letter 'L' on his keyboard and asked Duvall over the telephone line if the character had appeared on the SRI display. Duvall confirmed that the 'L' had come through. Kline then typed the letter 'O' and asked again over the headset. Duvall confirmed receipt of the 'O'.
Kline then struck the letter 'G'. Instead of displaying the character, the SRI host system suffered a memory overflow and crashed. The buffer on the receiving computer had been overwhelmed by the transmission, halting the session before the rest of the word could be entered. For roughly an hour, the historic digital record between the two nodes stood at just two characters: 'LO'.
After debugging the memory issue and resetting the software at SRI, the team re-established the connection later that night. Around 11:15 PM, Kline typed out the full five-letter command 'LOGIN' without interruption. The remote session succeeded, marking the first complete host-to-host transmission over a packet-switched computer network.
The Logic of Packet Switching
The significance of the 1969 test lay not in the letters sent, but in the underlying transmission architecture. Traditional telecommunications relied on circuit switching, the technology used by the telephone network. In a circuit-switched network, a dedicated physical or virtual circuit must be established and reserved for the entire duration of a communication session, regardless of whether data is actively traveling across the wire. This model was highly inefficient for bursty computer traffic, where machines frequently pause between transmissions.
ARPANET was designed to validate packet switching, an alternative concept developed through independent research by Paul Baran and Donald Davies, along with theoretical foundations in queueing systems explored by Leonard Kleinrock. Packet switching breaks digital messages down into small, standardized chunks called packets. Each packet contains payload data along with header information specifying its source, destination, and sequence. These packets travel independently across shared communication links, allowing multiple computers to share network capacity dynamically without reserving private lines.
The Interface Message Processor
To handle the routing and transmission of these data packets, ARPA contracted the engineering firm Bolt, Beranek and Newman (BBN) of Cambridge, Massachusetts. BBN designed and built specialized intermediate computers called Interface Message Processors, or IMPs. The IMPs served as the functional ancestors of modern network routers.
Instead of requiring every university mainframe to understand every other mainframe's unique operating system and architecture, each host computer only had to interface with its local IMP. The IMPs handled packet assembly, error checking, routing, and delivery across dedicated 50-kilobit-per-second leased telephone lines. The machine at UCLA was IMP Number 1, delivered in August 1969; the machine at SRI was IMP Number 2, delivered in October. The connection between these two machines formed the first functional segment of the network.
Building the Four-Node Foundation
The UCLA-to-SRI link was only the first phase of the initial network rollout. By the end of 1969, two additional research institutions were connected to ARPANET: the University of California, Santa Barbara, and the University of Utah. Each site brought distinct hardware and research focus to the project.
At UC Santa Barbara, researchers focused on interactive computer graphics and specialized mathematical systems on an IBM 360/75 host. At the University of Utah, computer graphics pioneer Ivan Sutherland and his team connected a DEC PDP-10. By connecting four fundamentally incompatible computer architectures across three states, ARPANET proved that distributed, heterogeneous computing networks were technically viable and resilient.
From ARPANET to the Modern Internet
Although the accidental 'LO' became an enduring piece of internet folklore—often celebrated for its resemblance to the archaic word 'lo', as in 'lo and behold'—the participants at the time viewed the event purely as routine debugging. The early network relied on early communication software known as the Network Control Program (NCP) to manage data flow between hosts.
As the network expanded to include satellite links, radio networks, and overseas nodes, NCP proved too limited for inter-network communication. This led to the development of the Transmission Control Protocol and Internet Protocol (TCP/IP) suite during the 1970s. On January 1, 1983, ARPANET formally adopted TCP/IP, creating the standard protocol framework that enabled disparate networks around the world to interconnect, forming the foundation of the modern internet.
Key takeaways
•The first transmission over ARPANET on October 29, 1969, was truncated to 'LO' when the receiving computer at Stanford Research Institute crashed on the third letter of 'LOGIN'.
•ARPANET demonstrated the practical viability of packet switching, which broke communications into discrete packets to share network lines far more efficiently than traditional circuit switching.
•Hardware called Interface Message Processors (IMPs), developed by Bolt, Beranek and Newman, handled network routing and allowed incompatible mainframe computers to communicate.
•The initial four-node network—comprising UCLA, SRI, UC Santa Barbara, and the University of Utah—formed the research foundation that eventually evolved into TCP/IP and the global internet.