The eye tracking explanation is a good one, thanks. I think it supports my hypothesis by providing a possible mechanism of action.

To your second point: yes, of course. But three weeks later it looks (mostly) the same but stops drawing compliments.