My wife and I recently moved half-way across the USA. Our new home is in a set of homes and the address contains a number sign, you know, this thing: "#", SHIFT-3 on my keyboards. I used the US Postal Service web site to file the address change and discovered that "#" is not allowed in the address field. Their web site wanted text, and suggested "APT". After trying several different combos, I ended up with APT, even though it wasn't correct. I figured that I'll be changing my important addresses myself, so this little discrepancy won't matter. Uh, that was my first wrong assumption.
Apparently the USPS is very efficient about notifying senders of address changes, and even my phoned-in updates to some important accounts were over-ridden by the USPS notification. So I've been painstakingly phoning, and in some cases writing letters, to make sure my mailing address is accurate. I updated an insurance policy address and just today needed to update again since "#" translated as "APT"; we had to settle on the word "UNIT".
What is it with these computer systems? As far as I can tell addresses can be very free form. My son's address includes the phrase "2nd floor". There are plenty of fractional street addresses; my new home town has a 38½ Street.
For me this is just another example of a poor system design and engineering. I'd love to see the use cases and requirements specifications for the content of the address field. Or was this just an ad-hoc decision made by the programmer while writing the code? Now that computers are here to stay, and we need to rely on them to get real work done, recognizing and accommodating the details of the lived, human world are more important than ever. These systems are already too hard to use. Why am I always adapting to the technology? When will it start adapting to me?