Re: I believe OCR is used for sorting the mail
Actually, of more interest to marketers. I used to work for a postal company, and the OCR machines we had were Siemens (yes) integrated mail processors (IMP). They could pretty much do everything. OCR readability was pretty good, I think the target was about 96-97% of all addresses.
The most interesting thing was that the machines took a photograph of the letter, not just the address and could store it. The main purpose of this was that if there was an address that could not be read, the image was then flashed to a VDU where a human could then input the address and send it back to the IMP, which would then route the letter to the correct sorting bin.
One of the crazier ideas was to create a database of every delivery point in the country, called NAD.. yes... National Address Database and assign an individual identifier for it and store the images for the letters going to each address. As company logos, etc, are often printed (and could be read by the OCR), it would eventually have built up a database on every household that a marketer would give their right testicle/ovary for. We would have known who you banked with, had insurance with, what loyalty programs you were a part of, who supplied your telecoms, power etc...
Fortunately it didn't go ahead back then as storing and sorting that data on a daily basis was going to be very expensive....now, I don't know, the concept would be even easier and cheaper to implement.
It's the not the spymasters, it is the marketers I fear.