You're onto something here - and Google has already digitized many, many books. This has enormous potential value as a collaborative endeavor; Google might not object if they were made a party to it and the search conducted at a low intensity. Alternatively, some kind of peer-based comparison a la BOINC ought to be possible. It has value beyond simple plagiarism detection.

Why do you think that would be against Google's ToS? I wouldn't guess so.