I know that if you hit the "suggestions" button near the "Up Next " it gives you some machine learned suggestions for books to read and the rationale behind the suggestions. I think that's great.

What I would love is for one of the outputs to be a book that is extremely different (but still "good") from everything else I read. It should be able to do his by parsing that same data that it uses to select books based on what you have been reading, and what fits with your current reading trends and all that. This time though, it just finds the under represented/less logged moods, pace, type and genre from your data.

I feel like this would allow for more visibility into the corners and blindspots of our reading selections and allow us to broaden our reading horizons more easily.