I suggest crawling Goodreads for books and quotes, rather than crowdsourcing: you could instantly have 3 million books and quotes added to the site.

You might argue that this is against the idea of your site (you seem attached to the serendipity idea) but, if you are crowd-sourcing books and quotes, you'll end up getting something very similar to this with a lot more work. Look at the books and quotes you already have -- they are all in the top few percent of Goodreads, and the quotes have already been selected.

The point is not that you don't want to use algorithms; the point is that you want an algorithm that allows for serendipitous exploration.

All very easy to do! Am happy to do myself if you put the site up on github.