The greater problem of curation is that it doesn't scale, and you need immense human effort to survey and curate both the breadth of questions -- what's a good table saw? what aspects of Egyptian culture were exported back to Greece?
is HDPE plastic safe? give me some punk music. -- and also the breadth of answers, both every website and every type of table saw.
The lesser is that you cannot curate without introducing a voice, a set of preferences that may not be universal. Tastes are not universal, you can't recommend the same band for everyone. Resources are not universal, regardless of whether the $10000 table saw is more than 100x better than the $100 table saw, it's just out of reach of most people. And needs aren't universal -- a professional cabinet maker and a DIYer making a chicken coop don't need the same saw.
There's a set of priors behind every query, and you either need to get users to frame their queries in a way that captures all of the relevant priors, or you need to create a variety of voices that capture different sets of priors and curate answers appropriate to that voice. Are you asking Norm Abrams, Monica Mangin, or Shane Wighton for a recommendation on a table saw?
Perhaps there can be a difference between search engines for answering specific questions, and directories where one may browse a broad range of topics without any goal in particular.