Ok, the link disappeared as fast as it appeared, but the story went like this:

The SSA sent out letters to a number of recipients saying their benefits would end in 1900. These recipients were basically widows whose youngest child would be reaching the age of 18. SSA blames the problem on the software that generated the letter, insisting that their internal software is remediated.

This raises two questions:

What good is it if your software is working, if software you interface with isn't? Why did the link dissappear so fast?

-- Amy Leone (leoneamy@aol.com), September 08, 1999

Answers

Here's the address (sorry, don't know how to link in here) http://www.seattletimes.com/news/technology/html98/glit_19990904.html

This was a link posted on a message yesterday.

Boz

-- Boz (Boz_Inc@yahoo.com), September 08, 1999.

Anyone who's seen my posts will know that I'm no polly, that I'm looking for the big bump that will awaken the sheeple.

That being said, this probably isn't the bump.

It all depends on where the "1900" comes from. If the letter printing program has a hard-coded "19", and just adds the last two digits, then this is a cosmetic error, just as the SSA people have protested, and essentially trivial. BUT, if the entire date is passed to the letter printing program as four digits, and the date-computations within SSA core functions believe this person's eligibility expires in 1900, then we have a BIG problem.

We just don't know, yet. It's splashy and embarrassing and just what we were looking for, but it might be trivia.

-- bw (home@puget.sound), September 08, 1999.

...if the letters (only) were hard coded, then all letters would have 1900. Since the 'puter system picked up only 32,000 and mailed them, then calculations for people who would be dead? or whatever the factor to end benifits would be calculating dates of birthdays on January.

-- dw (y2k@outhere.com), September 08, 1999.

Not sure I'm following you, dw. If the "19" is hardcoded, then all letters would have "19xx", where xx is some year. You obviously don't hardcode "1900", so they wouldn't show that.

The 32,000 count might be some batch (the first run to print this particular letter with dates in 2000, and all the records in the batch have the same expiration month). The 32,000 might also be a *subset* of some batch (the first run that *could* include dates in the year 2000, the 32,000 being a small fraction of the run where the expire date is calc'd in 2000, the others calcing in 1999, etc).

Not enough information yet. We certainly don't know that the computer system picked out only 32,000 to be mailed in this run.

-- bw (home@puget.sound), September 08, 1999.