To be fair, the test is pretty biased as they only tested use cases that are bound to be better as physical controls. It would have been more interesting if they had includes cases like selecting a route on Google Maps or navigating an app like Spotify.