ARPANET and the First Message: How the Internet Began in 1969

On the evening of 29 October 1969, a student programmer at the University of California, Los Angeles typed the first characters of a message meant for a computer about 560 kilometres away. The system on the receiving end was at the Stanford Research Institute (SRI) in Menlo Park. The network between them was ARPANET, and it is the direct ancestor of the internet you are using today.

Why ARPANET was built

ARPANET was funded by the Advanced Research Projects Agency (ARPA) of the United States Department of Defense. In the 1960s, ARPA paid for computers at universities and research labs around the country, and each one had its own terminals, software and users. Researchers who wanted to use a machine elsewhere often had to sit at a different terminal for each one. The project manager Larry Roberts set out to connect these machines so that expensive computing resources could be shared.

A common myth says ARPANET was designed to survive a nuclear attack. The people involved have generally described the practical goal as resource sharing between research sites. Resilience was an interest of some earlier theorists, but it was not the main reason the network was funded.

Packet switching: the key idea

The old telephone system worked by circuit switching. When you placed a call, a dedicated path was reserved between the two ends for the whole conversation. ARPANET instead used packet switching: data is cut into small chunks, each chunk carries its destination, and the chunks travel through the network independently and are reassembled at the other end.

The idea developed in parallel in two places. Paul Baran at the RAND Corporation in the United States wrote about distributed communications and message blocks in the early 1960s. Donald Davies at the National Physical Laboratory in the United Kingdom independently developed similar ideas and introduced the word "packet". Leonard Kleinrock, whose lab at UCLA hosted the first node, had published early theoretical work on queueing in networks.

The IMP and the first four nodes

The network did not connect the research computers directly. Each site got a dedicated minicomputer called an Interface Message Processor, or IMP, built by the Cambridge, Massachusetts firm Bolt, Beranek and Newman (BBN). The IMP handled the network work, and the local host computer only had to talk to its IMP. This separation is an early example of layering, a design habit that runs through all later networking.

The first IMP was installed at UCLA at the start of September 1969. The second went to SRI, then came the University of California, Santa Barbara and the University of Utah. By the end of 1969, those four sites were connected.

The first message

The first attempt to log in from UCLA to the SRI machine was made by Charley Kline, working in Kleinrock's group, with a person at SRI on a phone line confirming what arrived. The plan was to type the word LOGIN. According to the accounts from UCLA, the system crashed after the first two letters, so the first message carried across ARPANET was "LO". The team fixed the problem and successfully sent the full login later that evening.

It is a nice story, but there is a practical point behind it: the first real use of the network was a test, it failed in part, and the engineers fixed it and tried again. That is how most infrastructure gets built.

What ARPANET grew into

The network expanded quickly during the early 1970s, adding university and research sites across the United States and then links to other countries. Several things appeared on it that people still use:

• Email, which appeared in 1971 and within a few years became the network's most popular service (see our article on Ray Tomlinson and the @ sign).
• File transfer and remote login protocols.
• Public demonstration at the International Conference on Computer Communication in 1972, where the network was shown to a wider audience.

The original host protocol, NCP, was eventually replaced by TCP/IP. ARPANET was formally shut down in 1990, by which time the wider internet it had inspired had long outgrown it.

Design lessons that still apply

ARPANET idea	Where you still see it
Packets routed independently	Every web page, email and video stream
Dedicated network processors at each site	Routers and network cards that offload work from your server
Shared expensive resources over a network	Shared hosting and cloud computing
Test, fail, fix, retry	Staging sites and deployment checks

How ARPANET compares with today's internet

ARPANET was tiny by modern standards. In its first years it connected a few dozen computers, and each link ran at speeds that a modern phone would consider laughably slow. What survives is the architecture rather than the scale: independent networks, agreed protocols and traffic that finds its own way. When a visitor opens your website, their request crosses several networks owned by different organisations, none of which has to know about the others in advance. A content delivery network adds one more layer to that path by serving copies of your pages from locations closer to the visitor.

It is also worth remembering how much of the early network was built by small teams of academics and engineers sharing their results in public documents. That culture of writing things down and publishing them, which later became the RFC series, is a large part of why the internet could grow without a single owner.

The practical lesson for site owners

ARPANET's real achievement was making a network that many different machines could join without any one of them being special. Your website depends on the same principle: it works because hundreds of independent networks cooperate. The part you control is the server your site lives on, so choose one that is reliable and well supported. If you want to see the options, our hosting plans run from shared hosting to dedicated servers, all with the Plesk control panel.