The gps thing is even more amazing that what he said in the article because it turns out that the receiver doesn’t need an accurate clock of its own for it to work
That's right, you get rid of the requirement of its own clock, and instead accept that you need to "see" one more satellite than you'd need otherwise.
The (oversimplified) mental model is this:
If the receiver knows the exact time, and sees a signal from one satellite, it knows its position in one spatial dimension.
For each satellite signal you add, you can determine the position in one more dimension. Once you get to the 3rd satellite, you have no more dimensions to add. Instead, the 4th satellite lets you drop the requirement for your own (absolute) time keeping.
After that, each satellite you see improves the overall precision.
If the receiver has some internal sensors that can also help (gyros of various kinds).
I remember as a kid my father was working on the first version of the Air Combat Maneuvering Range (this was back in the late 1960s.) This involved dogfights with simulated missiles and guns. To work, the system needed to accurately track the position of the aircraft. Radars on mountains around the range would give good positions in two dimensions, but poor vertical position (this axis was nearly perpendicular to the line from a radar to a plane). The solution was to add a pod to each aircraft with various gyros and incorporate this information via a Kalman filter. Nowadays such a system would just use GPS, much simpler.
On top of that, with modern surveying and an RTK GPS base station you can get centimeter accuracy for very precise measurements of land, all by carrying a stick with a receiver on it. Most populated counties in the US already have a network running so you usually just need the receiver.
