That inertia is indeed huge. For example, if your library uses RFID but also shares a catalog and their books with other libraries, they'll still need barcode labels (with the same number that's encoded on the tag) so that the other libraries can deal with the items. Barcode scanners are ubiquitous in libraries; RFID readers, not so much.
Also, one of the promises of RFID for libraries has not panned out very cleanly. When a library does an inventory, not only do they want to verify the existence of the books, but that they are _in order_ on the shelf. RFID vendors for libraries promised that you could do this "shelfreading" accurately just by passing the wand across each shelf, but for various reasons the results are imperfect enough that it's not a clear winner over doing it with a barcode scanner. Given the fact that RFID tags have historically been far more expensive than barcode labels, the economics don't pencil out for many libraries to switch to RFID.
Also, one of the promises of RFID for libraries has not panned out very cleanly. When a library does an inventory, not only do they want to verify the existence of the books, but that they are _in order_ on the shelf. RFID vendors for libraries promised that you could do this "shelfreading" accurately just by passing the wand across each shelf, but for various reasons the results are imperfect enough that it's not a clear winner over doing it with a barcode scanner. Given the fact that RFID tags have historically been far more expensive than barcode labels, the economics don't pencil out for many libraries to switch to RFID.