Before networks: one machine, many people
The earliest electronic computers of the 1940s and 1950s were standalone machines. To use one, a person brought a program, often on punched cards or paper tape, and waited for results. Moving data between machines meant carrying cards or reels of magnetic tape from one room to another. In that sense, the first “network” was people with boxes.
Even so, the idea of linking machines over distance appeared early. In the 1950s, large military projects such as the SAGE air defense system in the United States used telephone lines to carry radar-derived data to computers and operators in different places. Airline reservation systems developed around the same era, around the late 1950s and 1960s, gave agents in many cities access to one central computer. These systems were custom built for one purpose, but they showed that a computer’s usefulness grew when many people at remote locations could reach it.
Time-sharing and remote terminals
The first widely felt step was time-sharing, which took shape in the early 1960s. Instead of running one job at a time, a time-sharing system switched rapidly among many users, each at a keyboard-and-printer terminal. Terminals could sit in another building, or in another city, connected over ordinary telephone lines with devices called modems that convert digital signals into sound and back.
This arrangement is sometimes described as a network, though it was really a star: many dumb terminals connected to one smart machine. The terminals did little thinking of their own. Still, time-sharing created habits that later networking inherited, including user accounts, shared files and messages left for other users on the same system. Time-sharing also exposed the first security questions about who may see what, a theme covered in the story of early security problems.
Computers talking to computers
The next step was letting machines, not just terminals, exchange data on equal terms. Several groups worked on this in the 1960s, and credit is shared. Researchers in the United States and the United Kingdom independently explored the idea of cutting messages into small pieces, sent separately and reassembled at the far end. This approach, now called packet switching, suited bursts of computer traffic better than the telephone system’s dedicated circuits.
Government-funded research produced the ARPANET, whose first links came into service in 1969 among a handful of universities and research centers. Its designers wanted researchers to share scarce computing resources, and it quickly became popular for something less expected: electronic mail. Other experimental networks appeared in Europe and elsewhere in the 1970s, and commercial public data networks, often based on a standard called X.25, followed. By the late 1970s there were many networks, but each used its own rules.
From one machine to many: local networks
While wide area links stretched across countries, a different problem arose inside buildings. As minicomputers and, later, personal computers spread through offices and campuses in the 1970s and 1980s, people wanted to share printers, disk storage and files without a long cable run to a central machine. Ethernet, developed at Xerox PARC in the 1970s, became the best-known answer, though it competed with other designs such as Token Ring and ARCnet. Network software from several vendors let small offices share files and printers using inexpensive hardware.
Local networks changed how organizations bought computers. Instead of one expensive central machine, a company could connect many smaller ones. Later waves of the same idea led to the client-server model, where shared machines hold data and individual desktops request it. The 1980s saw this approach and the spread of home computing arrive more or less together.
Rules that fit together
By the early 1980s, a practical problem dominated: networks built by different makers could not talk to one another. Companies had their own network designs, and standards bodies proposed layered models in which each layer handles one job, from the physical cable up to the application. A separate set of protocols, TCP/IP, grew out of research on connecting dissimilar networks. The ARPANET adopted it at the start of 1983, and over the following years it became the common language of what we now call the internet. The long argument over competing designs is told in more detail in the history of network architecture.
The internet and the web
The 1980s also brought university networks, mailing lists, Usenet discussion groups and bulletin board systems run by hobbyists on ordinary phone lines. In the late 1980s and early 1990s, national academic backbones expanded, and restrictions on commercial use gradually eased. The World Wide Web, proposed around 1989 and 1990 by Tim Berners-Lee at CERN, added an easy way to publish and follow links. The web is a service that runs on the internet, not the same thing as the internet itself, a distinction that many readers find useful to keep straight.
Home users reached the network first through dial-up modems and later through faster broadband links. Wireless networking spread in the late 1990s and 2000s, and mobile phones turned the network into something carried in a pocket.
Trade-offs along the way
Each stage solved one problem and created another. Connecting machines made sharing easier, but it also made it possible to attack a machine from far away, so networking and computer security grew up together. Open, shared protocols encouraged growth but were designed in a more trusting era, and later systems have often had to add protections afterward. Growth also left gaps in who could get connected, a theme explored in work on the digital divide.
A rough timeline
- 1950s: military and airline systems link remote terminals to central computers.
- Early 1960s: time-sharing makes terminals common.
- Mid to late 1960s: packet switching is developed by several groups.
- 1969: the first ARPANET links go live.
- 1970s: Ethernet and other local networks appear; public data networks spread.
- 1983: ARPANET switches to TCP/IP.
- Late 1980s to 1990s: academic networks expand and commercial use grows.
- Around 1990: the web is created at CERN.
- Late 1990s onward: broadband, wireless and mobile access.
Dates in this field are often approximate because projects overlapped and early experiments were not always announced at once. The pattern is steadier than the dates: first shared machines, then shared lines, then shared rules, and finally a shared global network that most people now use without thinking about the layers beneath.