How a Napkin Sketch Became Ethernet
In 1973, Bob Metcalfe was trying to solve a practical problem at Xerox PARC: how could many computers share one communication channel without their transmissions repeatedly interfering? While staying at a friend’s house, he sketched an idea for a network on a napkin. It was a starting concept, not a finished cable specification.
The key inspiration came from ALOHAnet, a radio network developed in Hawaii. Because several stations used the same radio channel, two could transmit at once and garble each other’s messages. ALOHAnet’s approach was to try again after a collision rather than reserve the channel permanently for one sender.
Listen, Send, and Try Again
Metcalfe adapted that shared-channel idea for computers connected by wire. In early Ethernet, devices used a method called carrier sense multiple access with collision detection: a computer listened for activity before sending, monitored the channel while it transmitted, and stopped if it detected that another transmission had collided with its own. The devices then waited different amounts of time before retrying, reducing the chance they would collide again immediately.
That mechanism let multiple computers use one cable without a central controller assigning every turn. It also made Ethernet’s limits visible: as more devices crowded a shared cable, collisions and retries could slow communication. Later Ethernet networks largely avoided that problem by giving devices dedicated links through switches, which direct data between endpoints instead of making every computer compete for the same wire.
The napkin captured the organizing idea—shared access with a way to recover from collisions. Turning it into dependable equipment and a common standard took further engineering, but that idea helped make it possible for computers from different makers to communicate over the same kind of local network.
