Same. At work we are in the process of rewriting an application that sends FAXes to doctors for their signatures. The barcode is on the edge of the doc, rather than the coverpage, but otherwise, same. The barcode is used as less error prone OCR for a record number so the FAX can be associated with the right "file". The users requesting the FAX loop view the returned document when it comes back to verify it (or at least, they will, when I finish recreating that feature :-) ). HOWEVER, users DO NOT sift through all incoming documents with data for other people which is none of a given user's business. Having to do so would be not only an inconvenience, but a data privacy issue.
Plus, I would rather somebody else build doc ID OCR in a language other than Java so I don't have to build it myself. This is going to come up time and time again.