TURNING POINTS
Beyond the Nuclear Myth: Why ARPANET Was Actually Built
A Pentagon-backed effort to share costly university computers became the technical foundation for a network of networks.
ARPANET began as a practical effort to share costly research computers, not as a nuclear command network. Its lasting contribution was a set of technical rules that let independent networks communicate.
In 1966, Bob Taylor had three computer terminals in his Pentagon office, each connected to a different ARPA-funded research computer. The machines could not communicate with one another. [1][2]
That small inconvenience pointed to a larger problem. Computing power was expensive and specialized, and ARPA was funding advanced work at universities and research centers whose machines remained largely isolated. Taylor, director of ARPA’s Information Processing Techniques Office, secured about $1 million for an experimental network. Larry Roberts managed the effort to connect those incompatible computers. [1][2]
ARPANET emerged from the Cold War defense establishment, but its initial purpose was practical: share scarce computing resources among researchers.
Building a Network for Researchers
ARPA created its Information Processing Techniques Office in 1962 under J.C.R. Licklider, who promoted interactive computing and communication among distant machines. By the mid-1960s, the office supported laboratories with different systems and different areas of expertise. A network could let researchers use a program or computer at another site rather than require ARPA to purchase the same capability again. [1][2]
ARPA did not want the connected host computers to handle every networking task themselves. In December 1968, it hired Bolt Beranek & Newman of Cambridge, Massachusetts, to build dedicated intermediary machines called Interface Message Processors, or IMPs. Frank Heart led the BBN engineering team. The IMPs carried traffic between local host computers and the network. [2]
This separated the network’s job from the research work being done on the computers attached to it.
The Nuclear-War Story Leaves Out the Main Purpose
The familiar claim that the internet was built chiefly to preserve military command after a nuclear attack contains part of the background, but it does not describe ARPANET’s original mission. At RAND Corporation, Paul Baran studied distributed communications systems that could continue operating after parts of a network were destroyed. His work contributed important packet-switching ideas. [2][3]
Baran’s RAND study and ARPANET, however, were separate projects. ARPANET was intended to share computing resources among ARPA-supported institutions, not to build a nuclear-hardened command network. Its early sites reflected that purpose: universities and research centers rather than military command posts. [2][3]
Packet switching supplied a useful method. Instead of reserving a continuous circuit for one exchange, a network could divide information into smaller packets and send them over available routes. The approach made more efficient use of communications links and reduced reliance on a single fixed path. Roberts drew on packet-switching concepts developed independently by Baran and Donald Davies at Britain’s National Physical Laboratory. [1][2]
The First Message Was an Incomplete Login
At 10:30 p.m. on October 29, 1969, UCLA student programmer Charley Kline attempted to log in from UCLA’s SDS Sigma 7 computer to an SDS 940 machine at the Stanford Research Institute. Bill Duvall received the transmission at SRI’s Augmentation Research Center. [4]
Kline began typing “LOGIN.” The remote system received the L and the O, then crashed. A successful login followed later, but the unintended “LO” became the record of ARPANET’s first host-to-host exchange. [4]
UCLA’s node was the first on the network. By December, it had been joined by the Stanford Research Institute, the University of California, Santa Barbara, and the University of Utah. The four-node system connected institutions with different equipment and research strengths. [1][2]
The builders also needed a way to settle technical questions. In April 1969, Steve Crocker issued Request for Comments 1, beginning the RFC process for circulating and refining network protocols. [2]
TCP/IP Turned One Network Into a Network of Networks
ARPANET connected its participating hosts, but researchers in the 1970s were also working with packet-radio and satellite networks. Connecting those systems required rules that did not force every network to adopt the same internal design.
Robert Kahn and Vinton Cerf developed a protocol suite for that problem. Their 1974 paper, “A Protocol for Packet Network Intercommunication,” described TCP, a forerunner of TCP/IP. The larger idea was open-architecture networking: separate networks could communicate while retaining their own internal structures. [3][6]
On January 1, 1983, ARPANET permanently changed from its earlier Network Control Protocol to TCP/IP. The shift did not immediately make ARPANET a commercial service or put most Americans online. It did establish common rules for connecting distinct networks at their boundaries. [3][5]
Later that year, operational military communications moved to MILNET while ARPANET continued as a research network. ARPANET was officially decommissioned on February 28, 1990, as newer systems, including NSFNET and commercial providers, overtook its infrastructure. [1][2]
ARPANET did not become the consumer internet by itself. Its durable legacy was an approach: send information in packets, permit routes to change, and use shared protocols so independent networks can exchange data. What began as an effort to get more value from costly research computers helped make possible a network of networks rather than one centrally owned system.
Frequently Asked Questions
Who sent the first ARPANET message?
Charley Kline, a UCLA student programmer, sent it from UCLA’s SDS Sigma 7 computer on October 29, 1969. Bill Duvall received it at the Stanford Research Institute. Kline intended to send “LOGIN,” but the system failed after the letters “L-O.” [4]
What did Paul Baran have to do with ARPANET?
Baran did not run the ARPANET project. At RAND, he developed influential ideas about distributed, survivable communications and packet switching. ARPANET’s designers drew on packet-switching concepts, but ARPANET was funded to share computing resources among research institutions. [2][3]
Why was TCP/IP important?
TCP/IP provided common rules that allowed separate networks to communicate while retaining their own internal designs. That open-architecture approach made interconnection among many networks possible. [3][5][6]
TIMELINE
Key moments
Sources
- ARPANET | DARPA — Defense Advanced Research Projects Agency (DARPA)
- Defense Advanced Research Projects Agency, A History of the ARPANET: The First Decade — National Security Archive / Defense Technical Information Center
- A Brief History of the Internet & Related Networks — Internet Society
- The Day the Infant Internet Uttered its First Words — UCLA Computer Science Department / Leonard Kleinrock
- NIHF Inductee Vinton Cerf Invented Internet Transmission Protocols — National Inventors Hall of Fame
- Robert E. Kahn Laureate Biography — The Franklin Institute
RELATED STORIES