It's not just a matter of covering things. Barcodes have the benefit of being human-readable, offering easy identification even when reading equipment misbehaves. Also, whenever they're handled by library staff, they're taking at least a cursory glance at the book in question.
RFID tags are an entirely different beast. They can be placed in an unobtrusive way, and are thus subjecting the book to a more honour-based system, especially if combined with patron-operated "self-check" borrowing and returning units. Fraudulent users can manipulate the system to avoid having the book registered onto their own account with very little creative thinking.
Sloppy programming or human interface design can lead to other sources of error, including tags reprogrammed with errant information. Since there's little evidence making the error imminent the moment it happens, it's gradually deteriorating the credibility of the data base to a certain extent.
Also, we're gluing these hardware stickers with a limited life-time into media we're expecting to last way beyond the lifetime of the sticker. The sticker usually doesn't carry any additional, human-readable identification, so the moment it fails to answer the reading device, the book will have to be re-identified and re-tagged manually.
As to the original concern: even placing the sticker inside the book doesn't warrant that it doesn't cover information. I've seen archived catalogs of art where there was no white space inside or outside of the brochure, and the sticker covered a substantial part of both imaging and text.
Then there are privacy concerns. "Who borrows which book" is a question we're not really allowed to ask. "Which books get borrowed more frequently than others" would be a valid question, but any kind of "who" going beyond general user groups is going to be problematic.