TCP/IP and the Invention of a Network of Networks

A successful network was not yet an Internet

ARPANET proved that packet switching could connect distant computers, but DARPA also sponsored packet radio and satellite networks with different speeds, interfaces and assumptions. Bob Kahn’s problem was how to let these distinct networks communicate without redesigning each one around ARPANET.^[1]

That problem is the essence of internetworking: create a system in which independently engineered networks can participate in a larger whole.

Cerf and Kahn designed a protocol above individual networks

Vint Cerf and Bob Kahn began developing the Transmission Control Program in 1973. Computer History Museum’s account places their first paper in that year and connects the work to DARPA’s need to link ARPANET, packet radio and satellite networks.^[2]

The architecture treated each underlying network as something that could carry packets without dictating the design of all other networks.

Gateways connected unlike networks

Special systems forwarded packets from one network to another. The end hosts did not need every network to expose the same internal switching behavior.

End-to-end reliability moved toward the hosts

The emerging architecture avoided requiring every network to guarantee perfect delivery. End systems could detect loss and recover, allowing the internetwork layer to remain comparatively simple.

The original TCP later split into TCP and IP

Early designs combined routing across networks and reliable end-to-end delivery in one protocol. The architecture eventually separated Internet Protocol, responsible for addressing and forwarding datagrams, from Transmission Control Protocol, which provides an ordered reliable byte stream for applications that need one.^[3]

This split also made room for applications that did not require TCP’s reliability model.

A 1977 demonstration proved three networks could operate as one

Computer History Museum records a 1977 demonstration linking ARPANET, the packet radio network and SATNET through the evolving Internet protocols.^[2]

The test mattered because the architecture had to survive real differences among terrestrial lines, radio links and satellite communications rather than succeed only on paper.

The internetwork did not hide every difference

Underlying networks still had different packet sizes, delays and failure modes. IP’s job was to provide enough common structure for end-to-end communication without pretending those differences did not exist.

Layering allowed each network to evolve independently

A radio network could change its internal technology without forcing applications on distant Ethernet or satellite networks to be rewritten, as long as the Internet-layer contract remained intact.

TCP/IP became a defense standard before the ARPANET conversion

Internet Society’s history notes that TCP/IP was adopted as a U.S. defense standard in 1980, giving organizations time to prepare software and operations for migration.^[1]

Standards adoption alone was not enough; every active host and gateway had to implement interoperable software.

January 1, 1983 became the famous flag day

ARPANET’s transition from NCP to TCP/IP occurred on January 1, 1983. Internet Society describes the event as a coordinated switch in which hosts that had not migrated would lose ordinary connectivity.^[4]

The successful conversion demonstrated both technical maturity and a growing capacity for community-wide operational coordination.

BSD Unix helped spread the protocols through research computing

DARPA supported work to integrate TCP/IP into Berkeley Unix. Internet Society identifies the inclusion of Internet protocols in BSD as an important reason TCP/IP spread throughout the computer-science research community.^[1]

Protocol adoption was therefore connected to operating-system distribution: networking became easier to deploy when it arrived as part of a widely used software environment.

Why TCP/IP is the architecture behind the word Internet

TCP/IP did not make all networks identical. It created a common layer in which networks could remain different internally while participating in shared addressing and packet delivery.

Internet Society’s IANA history marks the 1983 TCP/IP transition as a major institutional moment in the Internet’s growth.^[5] The enduring design principle is pluralism: one interoperable internetwork built from many separately administered networks.

RESEARCH / PROVENANCE

Works Cited

5 SOURCES

1. 01
   Internet Society — A Brief History of the Internet ↗internetsociety.org
2. 02
   Computer History Museum — 1973: Inventing the Internet ↗computerhistory.org
3. 03
   Internet Society — Short History of the Internet ↗internetsociety.org
4. 04
   Internet Society — Final Report on TCP/IP Migration in 1983 ↗internetsociety.org
5. 05
   Internet Society — The History of IANA ↗internetsociety.org

CodeHistory is a living archive. Citations document the evidence used for this edition; later evidence may refine the account.