Örn Hansen wrote:
Old IBM mainframes, used to have consoles that operators sat and watched for messages. When an error message, or any other message, popped up on the screen it was in the form of F50347, with some message attached and you had to look up in a pile of manuals, to figure it out. Because the message, was really short. Once you looked into the manual, it often referred to another message F40527 (These numbers are made up, merely to look like the originals and have no real relevance to the original messages), and often you had to look into yet another manual to read that description.
Those manuals were written, for security through obscurity, and not because the machines were so complex. We all know, security through obscurity, is not really a working security ... just as we know that 'idiot proof' only produces better idiots. We could land somewhere in the middle though ... don't you think?
One reason for those short error codes, was limited memory. Every byte counted. Incidently, according to a book about IBM's early computers, back in the '50s, core memory cost about $1/byte!