Web of knowledge - how the internet started

Like the ‘space race’, the internet was a by-product of the Cold War. In the early 1960s, when the world was haunted by the vision of nuclear war, the United States was shielded by a radar-based early-warning system that depended on the best computers of the day – large mainframe machines that filled a room. As part of its drive to leapfrog Soviet technology, the US Department of Defense set up the Information Processing Technology Office (IPTO) at its Advanced Research Project Agency (ARPA). It was given the task of finding a way to link all these computers into a secure network – one that would continue to operate even if part of the network were destroyed by a nuclear attack. The scattered computers of the tracking system known as the Semi-Automatic Ground Environment (SAGE) were linked by modems, but the dedicated telephone lines that connected them were still vulnerable to attack.

Packet switching

In 1962, J.C.R. Licklider was appointed head of the IPTO. He proposed an ‘Intergalactic Computer Network’, with implications beyond defence needs. It was a premonition of the internet. But how could it be achieved? Attention turned to a technique devised simultaneously by researchers in the United States and Britain. By using ‘packet switching’ for transmission along telephone lines, each block of computer data could be treated like the electronic equivalent of a ‘packet’ in the mail. The destination address was attached to the packet at the sending computer; it then travelled to its destination along the quickest route, via a series of computer ‘nodes’. If a node could not forward the packet directly, it would select another route. One way or another, the package would reach its destination. Individual transmissions could also be broken up into a number of packets before despatch and, according to availability, take different routes before reassembly at the destination.

Packet switching made optimum use of transmission capacity; and it was secure, flexible and robust enough to survive partial obliteration of the network. In 1969, the theory was tested for the first time in a network of four nodes: computers at the University of California, Los Angeles (UCLA); the University of California, Santa Barbara; the Stanford Research Institute; and the University of Utah. The so-called ARPANET was designed primarily for the military. But such computer links also offered university researchers the chance to share their findings. Other major American academic and research institutions soon joined ARPANET, and by 1972 it had 37 nodes. Meanwhile, parallel networks were created, linking other communities of computers.

 

Users of these networks found them particularly useful because they could carry ‘electronic mail’. In 1971, Ray Tomlinson, an American programmer at a company in Massachusetts, sent the first such message within a network – between two computers in the same room, linked to the ARPANET. He was also the first to adopt the @ symbol in the address.

Email, as it became known, could be flicked between the new personal computers (PCs) and workstations that appeared from about 1975. Designed for use by one person, they were equipped with ‘visual display units’ (VDUs) – screens displaying text.

The early networks were a province of a highly specialised, technical world, operated by computing experts. One of their most pressing tasks was to refine the methods, or protocols, that prepared information for transmission and allowed other computers to receive it. The earliest protocols were designed to link computers within a network. The next challenge was to devise protocols that could link one network to other networks: an ‘internet’.

In 1974, Vinton Cerf, a computer scientist at Stanford University, proposed a radically new way of restructuring the protocols to link networks: the Transmission Control Protocol (TCP), which breaks up data packets for transmission and reassembles them; and the Internet Protocol (IP), which sees that the packets find their way to the right destination. After nearly a decade of refinement, on 1 January 1983 ARPANET switched to TCP/IP, and TCP/IP soon became the standard set of protocols for networks, and remains so today. The internet had arrived. Computing still seemed remote to the general public, but it was beginning to enter the office. By the end of the 1970s, electronic typewriters were being edged out by ‘word-processors’. These had a memory capacity that enabled pre-printing correction and, after 1979, spell-checking, and electronic storage of documents and addresses. In the 1980s, word-processors had VDUs that allowed the user to see the document they were preparing without looking down. It was only a short step from here to PCs, which performed all the word-processing functions and carried more ambitious programs to produce documents using ‘desktop publishing’ (DTP), a process previously requiring a typesetter and printing press.