A network of networks: no centre, many paths | |
| Type | Global packet-switched network |
|---|---|
| Predecessor | ARPANET (1969) |
| Core protocols | TCP/IP, adopted 1 January 1983 |
| Governance | Distributed; no central authority |
The Internet is a global system of interconnected computer networks that communicate using a shared set of protocols. It is not a network but a network of networks: independently owned and operated systems that agree only on how to pass data between one another. The World Wide Web, email and streaming media are applications that run on it, not the thing itself.
Three decisions account for most of what the Internet is and is not.
Packet switching was proposed independently by Paul Baran in the United States and Donald Davies in the United Kingdom in the early 1960s. ARPANET, funded by the US Advanced Research Projects Agency, connected its first four nodes in 1969.
The problem that produced the Internet proper was joining ARPANET to other, differently designed networks. Vinton Cerf and Robert Kahn's 1974 design for TCP handled this by asking the networks in between to do as little as possible.[1] ARPANET switched to TCP/IP on 1 January 1983, the date usually given as the Internet's birth.
The Domain Name System arrived in 1983, replacing a single manually distributed file of host names. The National Science Foundation's NSFNET became the backbone through the 1980s, carrying a policy restricting it to research use; commercial restrictions were lifted in the early 1990s, and the network opened to the public. The Web followed, and with it the Internet's move from a specialists' tool to general infrastructure.
The Internet has no centre and no owner. It consists of tens of thousands of independent networks, called autonomous systems, that exchange traffic under bilateral agreements: peering, where two networks swap traffic at no charge, and transit, where one pays another for onward delivery. The Border Gateway Protocol propagates the routes between them.
Certain coordination functions must be global, because names and numbers have to be unique. Addresses and domain names are allocated through the Internet Assigned Numbers Authority and the regional registries, and protocol standards are published as Requests for Comments by the Internet Engineering Task Force, an open body that works by rough consensus and running code.
| Application | Protocol | Introduced |
|---|---|---|
| SMTP | 1971 (ARPANET), standardised 1982 | |
| File transfer | FTP | 1971 |
| Domain names | DNS | 1983 |
| The Web | HTTP | 1991 |
| Real-time media | RTP and successors | 1996 |
The Internet lowered the cost of distributing information to near zero, which redistributed influence away from those who previously owned the means of distribution (publishers, broadcasters, record labels) and towards platforms that aggregate attention. The same property that lets a dissident publish without a printing press lets a fraud reach millions. Debates over moderation, surveillance, network neutrality and the concentration of platform power are all, at bottom, arguments about who gets to decide what a permissionless network carries.