In the Beginning, there Was No Networking

Developer News News Flash: Adobe Has iPhone Workaround Adobe's Flash 10.1 Goes Mobile (Minus iPhone) A Salute to Visionary CEOs

Who Invented the Internet?

If you're old enough to remember what computing was like in the late sixties, you know that it was so different from computing today as to be unrecognizable. (Sobering thought: there are teenagers driving cars now who were born after the IBM PC's debut in 1981!) Back in the late sixties:

• The typical computer was a mainframe. Its computing power and memory capacity might be the equivalent of some of today's pocket computers. There were some computers smaller than mainframes, such as DEC's PDP series, but there were no personal computers.
• The typical user interface was the punched card.
• Some very lucky users might have had access to a dumb terminal, such as a printing Teletype Model 33 with 110 baud communications capability. These terminals provided "time sharing" access to the computer's resources.
• There was no such thing as a local area network (LAN) and very little in the way of computer networking at all.

ARPA scientists continued work on computer networking theory, and in a 1967 meeting in Ann Arbor, Michigan, Wes Clark proposed the concept of using a common subnet of computers to carry out networking functions. Thus a variety of mainframe operating systems would need to learn to talk only to the common subnet standard, and the concept of a "front end" computer was born. ARPA's first front ends were "Interface Message Processors" or IMPs, designed by Bolt, Beranek and Newman starting in fall of 1968.

The research and design that went into the ARPA IMPs was fundamental to the science of computer networking. The designers were inventing a new concept in communications, called packet switching, whereby messages are broken up into small chunks, each of which is tagged with the address of its destination network node. The packets are sent along network paths alongside other packets destined for other nodes on the network. This architecture is commonplace now, but at the time, ARPA engineers were arguing over whether the concept was even workable.

A lot of the early work centered around "packet radio," in which a network is constructed of radio links, not land lines. Here the past is definitely prologue, with those early discussions influencing later developments such as cellular telephony and wireless LANs.

Data first flowed between ARPANET hosts on October 1, 1969, when computers at Stanford and UCLA exchanged data thanks to two IMPs and a dedicated telephone line. By 1970 the ARPANET consisted of four hosts. One of the people who measured performance on the new network was Vint Cerf, then at UCLA.

With only four ARPANET hosts by 1970, it's clear that network became a real entity in the seventies, not the sixties. But what about the Internet? Cerf himself makes clear that the Internet was born in the seventies: "The Internet didn't emerge, in concept, until about 1972... In early 1973, Bob [Kahn] raised with me the question of linking SATNET, PRNET and ARPANET together into an internet. He called it the 'internetting' project and I took that on as a project at Stanford University where I had gone." (From "50 Questions with Vint Cerf," http://www.wiredguru.com/cd2.html)

Here, Cerf is using the term "internet" in its most fundamental sense: a network of networks. The ARPANET was a network of host computers. As computers became smaller and began to proliferate, a variety of networks began to evolve. Kahn and Cerf's fundamental breakthrough was to realize that the world could not afford to run a leased telephone line between every computer that might want to connect to every other computer. Instead, local or campus area networks would stitch together machines in a building or a campus, and "internetting" technologies would allow these networks to interconnect.

Kahn and Cerf collaborated on the creation of the Internet Protocol, or IP, and the Transmission Control Protocol, or TCP. Cerf continues: "[Details] of Internet started in 1974 and implementations in 1975.... By 1978, the specs and test implementations were stabilized and finalized." In 1981, TCP/IP began to outstrip the original ARPANET protocol, NCP, and on January 1, 1983, the Internet accepted only TCP/IP for data communications.

Original ARPANET applications protocols, such as Telnet (for remote login) and FTP (for file transfer) and SMTP (for email exchange) all were re-implemented for the Internet.