Unfortunately that’s only half good. It’s more typical to measure peripheral latency by connecting an LED to the switch to see when it’s actuated. This person performed an Apple 2s to oranges measurement by not controlling his keypress speed and not testing the Apple 2. The conclusions are based on feelings. Keyboard electronics are simple so short of severe mistakes they add negligible latency. HIDs are connected to interrupts so the OS should get the input pronto. The insanely large software stack will slow things down, but it isn’t the keyboard’s fault.

Edit: to exemplify the issue with this test: a poorly implemented capacitive touch keyboard would add 10 ms of latency and feel sluggish yet would be at the top of this chart.

Maybe he means key travel time?

I am also baffled by that comment.

The latency on the magic keyboard is only 15ms, as opposed to the common 30-40ms on other keyboards.

Key travel time contributes to this, so if you have longer travel you will also have longer latency (latency being the beginning of the keypress untill the character appears on your screen). I personally much prefer the feel of a mechanical keyboard, so I dislike both the magic keyboard and regular dome keyboards. The typing experience on a mechanical keyboard is better, even though it has more travel.