E.g. at a company we had to add "useless" content on our shop pages, otherwise google ranked the pages lower due to having not enough content.
If any feature/change introduces a regression, they will not ship it. (That kinda indicates that people outside of hn actually like amp)
Keep in mind that: amp team != search team.
Doubt. Especially "time spent on website" primarily indicates that the user didn't quickly find what they were looking for.
You'd be correct in assuming that it's a good indicator if it wasn't gamed. Since people believe that it's a ranking factor, they'll do whatever they can to increase it. Shitty SEO texts that go on and on without actually saying anything, breaking the back button, adding exit intent popovers, you name it, it's out there. All heavily increasing time on site. Are the users happy? Do you prefer a 5 sentence page that just plainly answers your question about some topic, or do you want a twenty paragraph novel that maybe, possibly somewhere includes the answer, but you need to completely read this opus magnum first?
Is that why answers are always buried in long articles or walls of text?
Seems a horrible metric.
Publishers definitely use it more heavily.