Ten years ago, I was a software engineer at the National Radio Astronomy Observatory [0]. We used the Very Long Baseline Array telescope to calibrate the rate a

Ten years ago, I was a software engineer at the National Radio Astronomy Observatory [0]. We used the Very Long Baseline Array telescope to calibrate the rate at which the Earth's surface moves, relative to distant pulsars. Atomic "clocks" are more like CPU "clock speed". Atomic clocks give a very precise measure of a time interval. But they don't tell you what time it is. The time of day needs to express both the idea of the rate of time, and also the position of the Sun and stars, relative to the position of the Earth. Like, we use "number of seconds since midnight, 1 January 1970" and expect that number to map to a time like "2018-12-25 20:37:44 UTC", and we expect that value to map onto a time about four days after the Northern Hemisphere winter solstice. Pulsar surveys make sure that those two concepts line up. Most of the time. [0]: https://nrao.edu https://nrao.edu