Random comments as I'm rushing through, so pardon the lack of context...
- You're right that we only see the people who survive, but I'm thinking more in terms of general web usage statistics. I'm also assuming that there are very, very few people who are interested in seeing only one specific book. Yes, we'd lose such people, but I imagine that others are here looking to read and'll try alternate searches. If there's no "Batman," then what about...?
- Google searches are easy to set up, but require someone to put the box on the page, and...I don't know, I've never been happy with the results. In this case, it'll be worse than what we already have, because it'll ignore the file names.
- I should retract my use of the word "technology," earlier, as I meant it semi-ironically. Technical adeptness has no more to do with navigating a web page than driving a car. It's all about conventions, expectations, and psychology.
Sadly, I can see the "right" architecture to make the searches turn things up correctly, but it ain't pretty. Among the things you want to take into account are the title (including variant spellings--U.S. Jones? US Jones? U S Jones?), the publisher (Rewl? Rural? Croydon?), and likely OTHER publishers (Holyoke or the original rights owners?). And then, once you've gotten that down, why not also index by contents, including characters and creators?
I wonder if that's what Zog was thinking about when he suggested setting up a wiki. Because if every book had a (crosslinked) page, each page could link to the direct download page, making it a far more effective search-and-browse mechanism.