ARPANET / The start of the internet
Can computers talk to each other?
That time, that place
A 1960s computer was the size of a room and the price of a building. Universities had one apiece at best, and the wish to use them from afar grew loud. A research arm of the US defence department funded a network. The question was how to connect them.
Why this question
Communication then meant the telephone model: seize one circuit between two parties and hold it, unusable by anyone else, until the call ends. Computer conversation comes in bursts with long silences, so a costly line would idle most of the time.
A different idea emerged: chop the message into small parcels, address each one, and let every parcel find its own way, to be reassembled in order on arrival. Instead of chartering one road, everyone shares all the roads.
What was found
On the night of 29 October 1969, a computer in Los Angeles tried to log in to another 600 kilometres away. The operator began typing LOGIN — and after the L and the O went through, the far machine crashed.
So the internet's first word was **LO**. An hour or so later the rest followed. Before the year was out four computers were joined this way, and the network was named ARPANET.
Networks before this had centres: lose the exchange and everything beneath it goes dark. The parcel design had no centre from the start — block one road and the parcels detour.
The live link proved conversation possible without chartered circuits. Even the two-letter crash was evidence of a kind: one machine died, the network did not, and the session simply resumed after repairs.
As networks multiplied, they needed a common agreement to join network to network; in 1983 that agreement, TCP/IP, became standard. 'Internet' means 'between networks'. In the 1990s the web — documents linked across it — was laid on top, and the present took shape.