It's even more impossible to satisfy them in a way that's also useful for Google's own business model.
The most obvious and potentially naive way of doing that would be to allow end users to upvote or downvote search results. I know that Google already does that in an automated way to some extend, but the problem is that that signal is then used to determine the quality across all the users which, as I mentioned in the beginning, is impossible to get right.
Instead that metric should only affect each user's own search results and not everyone else's. This could improve the quality of the results, bring more people back, and eventually increase the revenue. It would also help prevent "gaming" the system.
What am I missing here? I can't believe that they're not aware of that. I also can't believe that they don't want to fix it or that they're so focused on ai that internal politics don't allow them to do anything else. So what gives?