Bill,
"Why haven't we seen these dire Y2K consequences TODAY - since computers are currently calculating things like 30 year mortgages, bond maturities, financial projections,etc. etc. etc. NOW?!?"
The 30 year mortgage is a red herring for several reasons: (1) reporters seem to have picked up on it as an easily understood aspect of Y2K & spread it so that it has become urban myth; (2) such calculations would very likely be performed in a sub-sub-system; (3) when 30 year mortgages started hitting the wall in 1970 the sub-systems were fixed then; and (4) interest calculations typically use an integer, N (number of periods), not dates.
Long lived systems get repaired reactively as they break, not proactively.
A typical scenario is when Report X has been run on its regular quarterly basis & sent out to whomever (auditors, regulatory agency, etc.), the whomever calls back & asks why the report is WRONG. Someone looks at it & says OH &^%*! Then you scramble around, determine what got screwed up, determine how to rerun 6+ months of back production (remember you're still doing your daily production activity & your machine is running at 90% capacity). Activity like this is formally call "FIREFIGHTING"... it's a real macho thing to do, particularly if you enjoy getting phone calls at 2am.
A LOT of this sort of firefighting goes on day in & day out. But when it rises to abnormal levels, as it will in unattended Y2K scenarios, things (systems, people, etc.) will eventually break. People can only work so many 16 hour days in a row.
Not to forget that people will be attempting to repair systems that often have been running in minimal maintenance mode for many years. What, why & how the system works has been lost in the mists of time as the originators (both programmers & managers) have moved onto greener pastures.
Single example: I've just been sent some real numbers from my favorite "exemplary" site. These folks have a portfolio of 2.6M LoC in COBOL (40%) and two other 4GLs. They're in the "warehouse" business & have annual turnover (sales) in the US$700M+ range. These guys are very, very good. Far better than 99% of other mainframe shops.
They're about 50% done (they think) & are looking at a total bill of approx US$1.2M. This doesn't touch their desktop systems.
So extrapolate... this is just one IBM mainframe site, in the top 1% of effectiveness (by my own estimation from thousands of conversations with other mainframe shops).
I still hold that Y2K is A VERY, VERY BIG DEAL!
David |
