The traditional publishers provided two great services to the reader. They produced a book and set it out for sale. They said of the book: "A competent and trustworthy person has read this book and pronounced it better than competent and possibly interesting to you."
Granted, sometimes a book is published on name recognition or as money-laundering for services rendered by a politician, and the publishers' ambition to change the world has changed their definition of "competent" to favor "woke" checklists over story-telling or accuracy. But that imprimatur was, once upon a time, one of their great contributions to readers.
In our day, anybody can self-publish without anyone inspecting the work. Amazon has millions available for sale. How can I, as a book reader, sift through that? No one person can read even snippets of all of just the sci-fi out there, much less all the rest. And as a book writer, how do I get the word out about my eventually-to-be-classic work? A million others are shouting too.
Any solution will demand both trust and a team. I have to trust that the team member who actually read the book is competent to recognize the kind of quality I want, and classify it in a way that will make it easy for me to identify it as a candidate. That demands somebody to oversee the team and keep them up to the job.
Let's try a few back-of-the-envelope estimates. Suppose I want genre fiction--fantasy academy is supposed to be hot right now. Say there are a couple thousand books now, and a couple of new ones added every day. Suppose a volunteer can read 4/week--maybe 7/week if he can say "yuck" and not finish one. With a 3-man team of volunteers, that's 4 months before all the books in just this little sub-genre are classified. I do not say "reviewed": reviews aren't trivial to write. This is just a 0-5 on plot, 0-5 on characters, 0-5 on genre-match, 0-5 on intangibles, 0-5 on whatever else you want to record: diversity, adherence to Buddhist principles, whatever. And whether the reviewer finished the book. And a thumbnail of what its's about.
That's not a panacea. Suppose you've a reader who likes Andre Norton and detests Hal Clement. In order to distinguish the two you need to know not just the books' qualtities, but be able to classify them into sub-categories--of which there will be hundreds. Maybe the reviewers can keep track of them, but readers can't. Perhaps a text snippet would help.
And, how do you know the reviewer is doing his job? The administrator needs to either read or re-assign random books to see if the resulting rating is comparable. More time.
So far, so bad--but hey, within 4 months the backlog is done and it would be easy to keep up with new fantasy academy stories. The database would grow, and readers could put their oars in (maybe), but on the whole it would be a solved problem.
OK, now try all of sci-fi. F and SF at amazon is over 70,000 books. Scale up. 35x more books, and a hundred new ones a day (WAG) means if you have a staff of 100 you could keep up with the new books; you'd need 200 to finish in 2 years. You're not going to keep a staff of 200 volunteers for 2 years, and in a group that size you're pretty much guaranteed to have somebody who'll be happy to give good reviews for money or other favors, or because he likes the author's politics. You'll burn out people and start getting crud.
So, up the ante. Pay them. Make it a subscription affair. Who pays? If the authors pay it becomes a tax on authors and doesn't have a direct incentive to keep the output useful. If readers pay--well, everybody wants stuff for free on the net, so that may not bring in as much as you hope, especially since somebody will leak results. How much will you need? With pro readers, can you get 14/week? Better say 600/year--some people write monstrously fat volumes. To finish in 4 years (I'm using different target times. This isn't going to happen, or pay for itself, overnight.) you'd need at about 30--to keep up with new books you'd need 60! Suppose they learn to recognize mediocre stuff quickly, and you only need 30 people all told. Your payroll is north of $1M/year if you pay peanuts--and IT and other kinds of financial expenses and...
How much will people pay for this service? As stated, not enough--and there'd be free riders to cut down on business even more. If you could couple it with some kind of "book of the month club"-like arrangement you might get more to join. The new BOMC had about 100,000 members back in 2017. (I looked at their current ad and was not overwhelmed with the selection.) Maybe?
FWIW, one feature that might be handy with a BOMC-like service, albeit somewhat privacy-poor, would be to let you look at the lists of categories and of books and specify Not-Interested for categories.
With a BOMC-like tie-in there's an incentive to push the pricier stuff and not necessarily the good stuff, so maybe that's not such an ideal situation either.