The internet's ancestor, ARPANET, went online in 1969 to let four university computers share scarce processing power, not to survive a nuclear strike. That nuclear-survivability idea came from a separate Cold War study by RAND researcher Paul Baran, which ARPA borrowed for engineering reasons years later. The two stories get merged so often that most people now believe the myth instead of the record.
What ARPA Actually Said It Was Building
In 1966, ARPA program director Bob Taylor pitched Pentagon leadership on a problem that had nothing to do with warfare: he had three separate computer terminals in his office, one for each ARPA-funded research site, and none of them could talk to each other. Linking them into one network would let researchers share software, data, and processing time across sites instead of duplicating expensive mainframes.
Charles Herzfeld, ARPA's director at the time, later put it bluntly: Taylor and his colleagues wanted to see if they could link computers and researchers together, and the project's military role was much less important. That's the founding motive on record, from the person who approved the funding.
Where the Nuclear War Myth Actually Comes From
The nuclear-survivability story is real, but it belongs to a different person and a different decade. Rand Corporation researcher Paul Baran, working under Air Force funding in the early 1960s, was studying how military command networks could keep functioning after a first strike. He proposed breaking messages into small units passed node to node, so no single point of failure could take the whole system down.
Baran's concept wasn't adopted by the Pentagon at the time. Independently, British researcher Donald Davies at the UK's National Physical Laboratory developed the same core idea for civilian data networks and coined the term that stuck: packet switching. ARPA's network designers later folded packet switching into ARPANET's architecture because it was efficient and robust engineering, years after Taylor's resource-sharing pitch had already secured funding.
The Night the Network Said "Lo" Instead of "Hello"
On October 29, 1969, UCLA graduate student Charley Kline sat down to remotely log into a computer at the Stanford Research Institute, 350 miles up the coast. He typed "L," then "O," and the system crashed before he could finish "LOGIN." Roughly an hour later, after SRI programmer Bill Duvall adjusted the receiving machine, Kline tried again and completed the connection successfully.
UCLA professor Leonard Kleinrock, who supervised the project and had spent years developing the mathematical theory behind packet-switched networks, later called the accidental "LO" more prophetic than any planned first message could have been, since it happened to spell the first half of "hello." The romanticized version of internet history usually skips the crash entirely.
Four Computers Became a Global Network Through a Different Problem
By 1971, ARPANET had grown to 15 nodes. But it still wasn't "the internet" in any modern sense; it was one closed network among several incompatible ones springing up elsewhere. The real leap came in 1974, when Bob Kahn and Vint Cerf, working under ARPA's Information Processing Techniques Office, published a design for a common protocol that could let separate, independently built networks exchange data with each other.
That protocol became TCP/IP. Once ARPANET adopted it on January 1, 1983, and other networks like the National Science Foundation's NSFNET followed, "network of networks" stopped being a research goal and became a literal description. ARPANET itself wasn't decommissioned until 1990, quietly retired after the thing it helped build had already outgrown it. The organizational side of that expansion mirrors patterns seen in how modern crawling architecture handles scale that no single system was originally built for.
Why the Funding-Source-Versus-Motive Confusion Keeps Spreading
Both halves of the popular story are individually true: ARPA was a Department of Defense agency, and packet switching was partly inspired by nuclear-attack survivability research. Collapsing "funded by the military" and "designed to survive a nuclear war" into a single sentence is an understandable shortcut, but it erases the actual chronology and gives credit for the founding idea to the wrong motive entirely.
This is less a quirk of internet history and more a pattern in how technical origin stories get simplified for retelling. The same collapsing-of-distinct-facts problem shows up in coverage of how today's AI hardware race gets framed, where funding source and design intent are often conflated the same way. Security infrastructure built decades after ARPANET, like modern passkey authentication, inherited some of that same packet-switched network logic without inheriting the myth around it.
What This Means for How You Read Tech History Now
Funding source and design motive are two different facts, and conflating them is exactly how technology folklore gets built. ARPA paid the bills because it was a defense agency with research money to spend, not because Cold War survivability was the network's founding purpose. If you want to check whether a widely repeated tech origin story holds up, ask a version of the question this article just answered: who funded it, and separately, who actually designed the mechanism, and why. Those two names are rarely the same person, and rarely working from the same motive.
Frequently Asked Questions
Was the internet built to survive a nuclear war?
No. ARPANET, the internet's ancestor, was funded to let four university computers share processing time and software. Packet switching, the technology associated with nuclear-attack survivability, came from a separate RAND study by Paul Baran and was adopted later for engineering reasons.
What was the first message sent over the internet's predecessor?
On October 29, 1969, UCLA student Charley Kline attempted to type "LOGIN" to a Stanford Research Institute computer, but the system crashed after just two letters. The literal first message transmitted was "LO."
Who actually invented packet switching?
Paul Baran at RAND Corporation and Donald Davies at the UK's National Physical Laboratory independently developed the concept in the early 1960s. Davies coined the term "packet switching," and ARPA's engineers incorporated the idea into ARPANET's design later in the decade.
When did ARPANET become "the internet"?
ARPANET adopted the TCP/IP protocol, developed by Bob Kahn and Vint Cerf, on January 1, 1983. That protocol let separate, independently built networks exchange data, which is what turned a single research network into a network of networks.
Why does ARPA's military funding get confused with a military design purpose?
Because both facts are true and both involve the military, they get merged into a single simplified narrative in retellings. The funding agency and the specific design motive behind packet switching were separate decisions made years apart by different people.
When was ARPANET officially shut down?
ARPANET was decommissioned in 1990, after larger successor networks including NSFNET had already absorbed most of its traffic and functions.