The story
Packet switching is a way of sending data by chopping a message into small pieces, called packets, labeling each with its destination, and letting every packet find its own way across a network before being reassembled at the far end. It was worked out in the early 1960s by two men who did not know of each other at first: Paul Baran at the RAND Corporation in California, who described a design in a series of reports ending in 1964, and Donald Davies at Britain's National Physical Laboratory, who developed the idea in 1965 and gave it the name 'packet'.
The alternative was the telephone system's circuit switching, which reserved a dedicated line for the entire length of a call. That wasted capacity when nobody was talking, and a single cut line ended the call. Baran, worried about how the United States could keep its communications working after a nuclear attack, proposed a mesh of many nodes where messages were broken into blocks and routed around damage. Davies came at it from computing, seeing that computers send short bursts of data and need to share lines efficiently. The mathematics of queues for such networks had been studied by Leonard Kleinrock, whose 1961 doctoral work at MIT helped show they would work.
AT&T, which ran the phone network, doubted the idea would work. But in 1967 the researcher Larry Roberts, planning the ARPANET for the U.S. Defense Department's research agency, learned of Davies' work and adopted the term and a faster design. The ARPANET sent its first message in October 1969 over 50-kilobit-per-second lines. Packet switching was later combined with the TCP/IP protocols in the 1970s and 1980s into the internet, and the same principle underlies Wi-Fi and mobile data. Although Baran's nuclear-survival motive is often repeated, the ARPANET itself was built mainly to let researchers share expensive computers.
Why it mattered
- The ARPANET, the first large packet network, began in 1969 and grew into the internet.
- Wi-Fi and mobile data networks carry traffic as packets, as does nearly all modern communication.
- Many users could share one network cheaply, making online services and the web economically possible.
- The design of a network with no single point of failure influenced later thinking about resilient systems.
Sources
- On distributed communications (Baran, RAND, 1964) RAND Corporation
- Packet Switching Engineering and Technology History Wiki
- Why do we call them internet packets? His name was Donald Davies Internet Hall of Fame
- Donald Davies Internet Society
Drafted by AI (Claude) from the sources above and checked against them by AI; no historian has reviewed this page. Spot a mistake? Suggest a correction. How pages are made
Community notes
Loading notes…