Latency as a physical quantity

node

The delay between sending and receiving, and the property of the internet that is most stubbornly geographic. Light in silica glass travels at roughly two-thirds of its speed in vacuum — around 200,000 km per second — which works out at approximately 5 microseconds of delay for every kilometre of fibre. That number cannot be engineered away: no amount of capacity, compression or caching changes the transit time of a photon, so latency between two continents is set by the length of the cable between them and by nothing else. The consequences are visible and commercial. A New York–London route of about 6,000 km has a floor around 60 milliseconds round trip, which is why the Hibernia Express cable of 2015 was built and priced on being a few hundred microseconds shorter than its neighbours; financial trading firms pay for those microseconds because a market's price is a piece of information that arrives. Latency is also why detours matter after a break: when the Red Sea cables were cut in September 2025, cloud operators told customers that traffic previously crossing the Middle East would take longer, which is not a fault report but a statement about distance.

A node is not a place. Drawing it on a map would assert something about the world that no stored fact supports.