Incidentally, Google is clearly aware of this, and willing to step into some hot (and unprofitable in the short term) waters to get to the book-stored knowledge. They also revived decent open-source OCR probably for this purpose.