Last year I decided to use a totally random method (after reading an interesting article linked here) to choose my next book among my huge reading list. Was weird and fun, since it let me read things I would have put off for a long time.
I've started using a semi-random method. I pick a category, take a random page, and then pick whatever book I want from the twenty books on that page. I've enjoyed it so far. It gets me to read books that I may have put off for a while, but because I'm picking the top one out of twenty they're books that I really want to read.
I did this during that time, with iOS games. Since I work as editor for a large-ish app review portal, I have tons of games. Some I like, some I want to try... But too little time. So, I rolled a dice (well, a virtual one) to know which page and which game to try when I was in a playing mood. Doing this I "discovered" several lovely games I had hidden in my folders, I also did some huge cleanup of games I didn't like or no longer enjoyed and discovered I like backgammon enough to pick it up as one of my favourite board games (long behind go, but backgammon is faster to pick up and play on mobile.)
My reading list is also very long but I'm trying to mix on books that I think I'll really really enjoy with books that are good for me (that I'll still enjoy, but not as much).
In IMDb it is interesting, as you can (roughly) decompose ratings in 1s (haters), 10s (lovers) and the rest (who actually rate it). But I would say that the last part is the most important, unless you want to get the hype.
one method that works extremely well for me is to pick a book based on Amazon's "what other customer's bought". I noticed that some of the best books I read appear in clusters that other customers/peers valued equally high