the vibes are right of the original post but i feel like we're doubling the dunning effect here. maybe i'm tripling it, but like, sound off in the comments computer historians, but this explanation is bullshit?
the spec does standardize a reference date of 1875, so diagnosis would be: iso 8601 doesn't have an epoch, cobol's datetime storage does, and it's a different one, maybe they're storing iso 8601 dates as numbers (?) and using its reference date as a default?
To be clear, ISO 8601 *does* have an epoch, but it's not year 1875, it's year 0. The 2000 revision decided to define this epoch in a secular way by not saying "year 0 is the year before some guy thought Jesus was born" but instead that year 0 is the year 1,875 years before the meter convention.
First payment year minus retirement age : 1940-65 = 1875. So someone might have come up with some convoluted conventions to squeeze out some storage in the 60s/70s/80s.
Comments
https://www.tondering.dk/claus/cal/iso8601.php
http://www.g1smd.freeuk.com/FTP/ISO8601/PDF/ISO8601-1988_Issue-19880615_Edition-01_Published.PDF
I assume they haven't finished replacing it with DB2, and even if they had, could've migrated those 0's over.