Internet users welcomed the New Year this year with a controversy that reaches to the roots of the Internet itself. January 1, 1983 was the day computer systems on the ARPANET were required to switch over to the TCP/IP protocol. This year marks the 20th anniversary of that event.

Several news stories appeared on the Internet before or on January 1, 2003 heralding January 1, 2003 as the twentieth birthday of the Internet. Other news stories questioned calling this date the birthday of the Internet. To have the date of the Internet’s birthday be the subject of a controversy is appropriate, given the nature and history of the Internet. In its early development, the Internet grew and flourished because researchers were encouraged to debate their differences. In this environment, collaborative work thrived.

The controversy over the birthday of the Internet is no trivial controversy. It is a controversy about the defining nature of the Internet. What is the Internet? This is a question that must be resolved to understand which aspects of the Internet’s development qualify for attention and celebration.

There are those who claim that the appropriate birthday of the Internet is Sept 1, 1969, when the ARPANET began. They propose that the essential nature of the Internet is large scale packet switching. In packet switching networks, messages are broken up into small segments called packets. Each packet is routed to its destination separately. Packets are later reassembled into messages and delivered to their destination.

In its original conception, the ARPANET was a large scale packet switching network spanning the U.S. It was under the control of one entity, the U.S. Department of Defense (DOD). It was planned as a means of providing for resource sharing among diverse computers, diverse operating systems, and the diverse users of these computers. It was not designed to communicate with dissimilar packet switching computer networks. When the ARPANET was originally designed, there was no plan to link the ARPANET to any other packet switching networks.

NCP (Network Control Protocol) was the protocol used on the early ARPANET. A protocol is a set of agreements that make communication possible between different entities. This protocol made it possible for users to send a message to a subnetwork, the IMP subnetwork. (Interface Message Processor). NCP would also make it possible to transmit the message from the receiving IMP to the receiver’s computer. NCP’s role was limited. It did not function as a communications protocol. The IMP subnetwork provided this functionality.

Once transported to IMP subnetwork, the message would be broken into packets, the packets would be routed to their destination, using flow control, error checking and so forth. Then the packets would be reassembled into messages, and they would be transported to the receiver’s host computer.

Packet switching is recognized by all as an important advance in computer and communications technology. The controversy is over whether packet switching is the defining characteristic of the Internet, or whether the defining aspect is the communication among different and dissimilar networks?

To explore this controversy it will be helpful to look back at two time periods of Internet history. By 1973, there were a number of countries planning or developing packet switching networks. In the U.S. the ARPANET was being developed. In Great Britain, there had been the development of the National Physical Laboratory network (NPL network). In France, Cyclades was being built. There were networking research projects in other countries. The important question being considered by networking researchers was how to interconnect these different networks.

Robert Kahn had taken over the packet satellite network program at the DOD’s ARPA/IPTO and was planning to create a packet radio network program. He recognized that it would be essential to interconnect these networks with the ARPANET. The ARPANET had a reliable subnetwork. Its host protocol NCP expected reliability. But a packet radio network, for example, is highly unreliable. So there was the need for a new communications protocol which would supply the reliability that the IMP subnetwork had formerly provided. In developing the packet satellite network, researchers from Norway and Great Britain agreed to collaborate with U.S. researchers to develop protocols to connect these dissimilar networks.

In September 1973, Robert Kahn and Vint Cerf presented a design for a protocol for packet network intercommunication to researchers from several countries meeting at the University of Sussex in Brighton, UK. For the next 10 years there was continuing research and collaboration by researchers to create working implementations of this protocol for different computers and different operating systems.

From 1973 until 1982, researchers from Great Britain and Norway collaborated with US researchers using the ARPANET. By 1982, the Norwegian and British researchers could connect to the U.S. network using the internetworking protocol which came to be called TCP/IP (Transport Control Protocol/Internet Protocol).

The second time period is in 1983, on January 1, 1983, a cutover to the TCP/IP protocol was required by the U.S. Department of Defense for computers acting as host computers on the ARPANET. While not all the different computers could cutover by this date to TCP/IP, many did. By the summer of 1983, the cutover had been completed. The TCP/IP protocol enabled the host computers on the ARPANET to communicate with TCP/IP enabled host computers on other networks.

Other networks would differ technically from the ARPANET. To transport packets among these different networks meant that a whole new set of issues had to be understood and resolved. For example, packets on different networks would be of different sizes, there would be different decisions made regarding timing, flow control, error checking and so forth. There would need to be a means of having all the different networks recognize how to route packets to their destination address. A form of addressing was needed which would be recognized by all the networks of the Internet.

Solving such problems made it possible to provide for communication among the users and the computers on different packet networks. I propose that the design and development of an implementation of an internetworking protocol, which is called TCP/IP, is the defining characteristic of the Internet.

January 1, 1983 was the date set for the cutover to this internetworking protocol, TCP/IP, on the ARPANET. On this date the basis was set for users and host computers on the ARPANET and other networks to be able to communicate with each other. This is indeed a worthy candidate for the birthday of the Internet.