It's an old thread but there's some new info. Someone seems to have either started out from the same point with a bit more info, or built on the work in this thread (there's no backstory or attribution) and come up with an alternative and compelling scheme for the later codes here:
Cap date decoder at Vintage Amps Forum
He had access to a number of units built in this period all of which gave codes with 0 or 1 in the second position, decoding reliably in the form Year-Month-Month instead of our proposed Year-Week-Week. All of the codes we had considered in this thread to be YWW also work as YMM, generating a more probable distribution, with the exception or Richard's N WI 7 13 which is invalid both as WWY (week 71) and as YMM (month 13). Given the data set offered in the linked thread, I'm inclined to think of this as an error or exception for the moment.
The post also contains data for 1970s Daly caps, and extension of the TCC scheme to RS-rebadged versions. As yet, we have no scheme for Dubilier...