Internet History: From ARPANET to the Web and Beyond

On 29 October 1969, a team at UCLA sat down in front of a computer and tried to send one word to a machine up the California coast at the Stanford Research Institute. The word was LOGIN. The system crashed after two letters, so what actually crossed the wire was LO.

That is how the thing in your pocket started, with a half-finished word and a crash. Four computers were connected by the end of that year. Billions of people are online today. The internet was not invented. It was assembled, link by link, by people who each solved one problem and handed the next one to someone else.

Three decisions made that assembly hold together. Routing was separated from transmission. Control was shared between independent networks instead of sitting in one place. The protocols were left open, so any operator in any country could join without asking permission. All three look obvious now. None of them was obvious in 1969.

Here is how it happened, from the first theory paper to the network of everyone.

What the internet is and where the name comes from

The internet is a network of networks. Millions of local networks, in homes, companies and universities, connect to one another and speak the same language of protocols. That shared language is TCP/IP, the layered model that moves the internet, and it is why your laptop in Madrid can reach a server in Singapore without a translator in between.

The name comes from internetworking, the interconnection of networks, and that word carries the whole idea. The driving aim was to make different networks understand each other, which is a far harder problem than stringing cable between two buildings.

Data networks already existed before the internet. Every manufacturer built its own, and none of them talked to the others. The breakthrough was a design that let them join forces without being torn out and replaced.

One confusion wastes more time than any other. The internet and the web are not the same thing. The internet is the infrastructure. The web is a service that runs on top of it, like email or streaming video, and you can use the internet without ever opening a web page.

Now picture the scale. In 1969 the network had four nodes. Today a large share of the world’s economy, its power grid, its hospitals and its public administration depends on it staying up. None of that was planned. It is what happens when you leave an architecture open on purpose and let anyone plug in.

The early years: from ARPA to ARPANET

The origin sits with ARPA, a research agency of the United States Department of Defense created in 1958, a year after the Soviet Union put Sputnik into orbit. ARPA was not built to make weapons. It was built to fund the best research in the country’s universities, and out of that money came a computer-network project that changed everything. In those years people were still learning what a computer could do, and machines such as ENIAC were the closest thing to one that existed.

Licklider’s «galactic network»

J.C.R. Licklider put the idea on paper in 1962: a network of interconnected computers and the social good it would bring. He half-jokingly called it the «galactic network». Licklider had come from MIT and ran ARPA’s information-processing office, and from that desk he convinced the people around him that the network was worth chasing. He did not design the machine. He set the goal. DARPA’s own timeline shows how far that bet travelled.

Packet switching

The blueprint came with packet switching. In 1961 Leonard Kleinrock published a paper at MIT on the theory of communication networks that pointed straight at the model. Across the Atlantic, Paul Baran at RAND and Donald Davies at the NPL reached the same answer by different roads. Their question was simple. Why reserve a whole line between two points, the way a phone call does, when you could chop the message into packets and let each node pass them along towards the destination?

The payoff is easy to state. A network that sends packets uses its links better and survives failures, because every packet can take another path when one falls. Cut a phone line and the call dies. Cut a packet route and the traffic goes around.

In 1965, Lawrence G. Roberts and Thomas Merrill linked a computer in Massachusetts to one in California over a telephone line. It was the first long-distance connection of its kind, and it taught a clear lesson: the idea worked, but circuit switching was too rigid to carry it. Roberts came out of that experiment convinced the future ran on packets. Two years later he published the plan for ARPANET.

The first nodes and the first message

The first link of ARPANET went live on 29 October 1969, between UCLA and the Stanford Research Institute. The first message anyone tried to send was «LOGIN», and the crash after two letters left them with «LO». It is a small story, and it says everything about those years. They moved forward by trying.

By the end of 1969 there were four nodes. In 1970 came NCP, the Network Control Protocol, the first host-to-host protocol, which made real applications possible. In 1972 the ICCC conference hosted a demonstration that won over the doubters, and that same year Ray Tomlinson sent the first networked email and picked the @ sign because it appeared in nobody’s name.

Email deserves its own line. It was the first major application the network had never been designed for, and it spread faster than anything the labs had built on purpose. Half a century later it is still one of the main reasons people go online at all.

TCP/IP: the protocol that joined the networks

ARPANET was growing, and it hit a ceiling. NCP was rigid and built only for that one network. Linking networks with completely different architectures, radio, satellite, cable, needed something else. Bob Kahn and Vint Cerf set it out in 1974 and called it the Transmission Control Protocol / Internet Protocol, or TCP/IP.

The design is written down in two founding documents, RFC 791 and RFC 793. IP handles addressing and routing packets between networks. TCP makes sure they all arrive, in order and without errors. Split the two and any network can serve as transport, while the hard work sits at the ends, in the computers, and the wires simply carry what they are told.

ARPANET switched to TCP/IP on 1 January 1983. Engineers still call it «flag day», and many treat it as the technical birthday of the internet as you know it. Why does that date matter so much? Because it explains how the network grew without an owner. Nobody controls it. The protocols are open and public, so any operator can connect, and no single company can switch the network off.

The 1980s piled on the pieces you now take for granted. The DNS arrived in 1983 and turned numbers into names you can remember. Ethernet and local area networks carried the connection inside companies and homes. Each addition made the network more useful to people who had never heard of a protocol.

Openness sounds like a slogan until you watch what it does. A university in Europe, a telecom operator in Asia and a home router in Latin America all talk to each other because they implement the same public specifications. When a technology tries to close that door and push a private format, the rest of the network simply routes around it, with no committee forcing the issue.

The leap into the wider world: the web and the rush to connect

For years the internet stayed inside universities and labs. Then the 1980s put a PC in the living room, and something that once sounded absurd started to make sense: one person, one computer, at home. Big state programmes such as NSFNET in the United States and JANET in the United Kingdom poured money into the network, built transoceanic circuits and opened it to more users. By 1990, when ARPANET was switched off, more than 100,000 computers were already connected worldwide.

1991: the World Wide Web is born

The next leap belongs to Tim Berners-Lee at CERN, Europe’s particle-physics laboratory. Scientists there had needed an easy way to share documents for years, and he built one, a system for storing and retrieving pages that he called the World Wide Web. On top of the internet that already existed he added three pieces: HTML to write the pages, HTTP to serve them and the URL to find them. Put those three together and you get something a non-engineer can use. The first website went live in 1991 and explained what the web was.

Notice what Berners-Lee did not do. He did not patent it, and CERN placed the specification in the public domain in 1993. Had the web been closed, it would have become one more proprietary service of the era. Instead it ran past every rival and became the face of the internet. The same open standards are still maintained by the W3C today.

Browsers and search engines

The web needed a window, and the window was the browser. The first one to catch on was Mosaic in 1993, built at the NCSA by Marc Andreessen and Eric Bina. A year later Andreessen founded Netscape and shipped Netscape Navigator. Text turned into images, links and forms. As the number of pages exploded, search engines became the only way to find anything, and email went from a tool for a few to a daily habit for millions.

That trio, pages, a way to read them and a way to find them, is what turned the network into something ordinary. Before the web you had to know how to operate the internet. After it, you could just use it. The pattern repeated with social networks, with streaming video and with the phone.

The dot-com bubble

The enthusiasm had a hangover. Around 2000 the dot-com bubble burst, and a crowd of companies that promised to reinvent commerce without a business model went under. The network did not sink. It came out stronger, with real users, solid infrastructure and proof that this had staying power. Instead of fading, it set the stage for the decade that followed: social networks, streaming video and the smartphone as the front door.

Timeline of the history of the internet

The whole run, from the first theory paper to the network of everyone, fits in a single table.

Year	Milestone	Why it matters
1961	Kleinrock publishes the theory of packet switching	The technical idea that makes the network possible
1962	Licklider describes the «galactic network»	Sets the goal before the blueprint exists
1965	Roberts and Merrill link two computers thousands of kilometres apart over a telephone line	Proves circuit switching is too rigid
1969	ARPANET is born: first UCLA–SRI link and four nodes by year end	The embryo of the internet
1971	Ray Tomlinson sends the first email with the @ sign	The first mass application of the network
1974	Cerf and Kahn formulate TCP/IP	A common language for different networks
1-1-1983	ARPANET adopts TCP/IP, «flag day»	The technical birth of the internet as we know it
1983	The DNS is created, the domain name system	Names instead of numbers for every machine
1990	ARPANET is switched off and more than 100,000 computers are connected	The network outgrows its original scope
1991	Tim Berners-Lee publishes the World Wide Web at CERN	The internet becomes easy to use
1993	Mosaic popularises the graphical web	Millions of people start browsing
2000	The dot-com bubble bursts	The network survives the hype and matures

The future of the internet

The history of the internet has been explosive. In a few decades it moved from a research programme to what is probably the most influential technology in history, and it has not stopped. The next stage points to a network that does not wait for you to type a request. It understands plain language and acts on your behalf. That is roughly what people call Web 4.0, pushed by artificial intelligence, by agents that chain tasks together and by the flood of connected devices we group under the Internet of Things.

Nobody knows what comes next. What is certain is that the internet is now critical infrastructure. Communications, commerce, energy and healthcare depend on it, which also makes it a target. The open design that guaranteed its success is the same property that makes it hard to defend. The door that lets anyone in is the door anyone can try to force.

The bottom line

The history of the internet fits in a handful of chapters and still changed almost everything. It started as a research bet to connect four computers. Packet switching made it possible. TCP/IP let it grow without an owner. The web and the browser made it usable, and the moment it became usable it stopped belonging to the experts and became the place where you work, learn and live.

Knowing where it came from changes how you see what it is. A network built on open decisions behaves a certain way, and that behaviour is the ground you stand on every time you build something on top of it. If you want the layer underneath, walk one step further back and read the history of computing. It starts much earlier, with the Jacquard loom and the Turing machine.

Frequently asked questions about the history of the internet

When was the internet born?

It depends on the criteria. ARPANET, its direct predecessor, brought up its first link in 1969. But the technical birth of the internet as a network of networks is placed on 1 January 1983, when ARPANET adopted TCP/IP and any network could begin to interconnect with the others.

Who invented the internet?

There is no single inventor. The idea was formulated by Licklider, the theoretical basis was supplied by Kleinrock with packet switching, the first link was made by Roberts and his colleagues, and the protocol that made it scale was designed by Bob Kahn and Vint Cerf. The internet is a collective work spread over decades.

What is the difference between the internet and the web?

The internet is the infrastructure, the network of networks with its protocols. The web is a service that runs on top of it, invented by Tim Berners-Lee in 1991. You can use the internet without opening a web page, for example with email, which is why the web is not a synonym for the internet.

What was ARPANET?

The experimental network funded by the ARPA research agency in the United States. It connected its first nodes in 1969 and was the direct predecessor of the internet. It was decommissioned in 1990, when the network had already grown far beyond its original scope.

Did the internet change with mobile phones?

Yes, a great deal. The smartphone multiplied the number of users and changed how people get online: from a desktop screen to one you carry with you. Today most traffic comes from mobile devices, and that has shaped the design of everything that is put on the network.