ISSUE 1677 Tuesday 28 December 1999

Worldwide Millennium chaos, and still four days to go By Robert Uhlig, Technology Correspondent

ONE of the fallacies surrounding the Millennium bug is that it will strike precisely at midnight on Dec 31, but it has already caused mishaps, confusion and failures several hundreds of times around the world.

Many disruptions occurred when computers or electronic systems infected with the bug tried to make a date-dependent calculation that ended in 2000.

Others took place because self-checking routines in computer software became confused when they attempted to forecast ahead and had to deal with a two-digit year date, such as 00 or 01 for 2000 and 2001, that is a lower number than the current year.

But most failures have been the result of human errors introduced when computers and electronic systems were being fixed and tested for the Millennium bug. In some cases, such as the troubles at the Passport Agency in August, the Millennium bug fixing process caused an already over-loaded system to collapse.

Most Millennium-bug-related mishaps have been covered up or averted before they caused noticeable damage. But among those that have been recorded are:

Last week, Wells Fargo, a large American bank, sent out certificate of deposit renewal notices to its customers dated Jan 1, 1900. The previous day about 3,600 customers in Iowa City received water bills that threatened them with an extra charge for payments made after Jan 3, 1900.

Earlier this month, when the Bermudan Stock Exchange went back to using an old computer system during a Millennium bug updating programme, its daily electronic exchange report listed dividend payout dates as 1900.

On Dec 9, Deutsche Bank was hit by a day-long computer shutdown that prevented it from undertaking international transactions for a day. A Deutsche Bank source told the trade publication Computer Weekly that the crash was related to the Millennium bug and happened when the bank was upgrading transaction management software.

In September, the Inland Revenue sent out National Insurance certificates which erroneously had dates printed in early 1900, rather than early 2000. A tax Certificate of Exemption stated: "You do not have to pay Class 2 National Insurance contributions from April 10, 1999 to April 6, 1902". Computer experts said it was almost certainly a problem with the Inland Revenue's mainframe systems, which came on top of other problems the Revenue has had while updating its systems for the Millennium.

In August, the Revenue erroneously threatened a law-abiding company with the bailiffs because Millennium bug work prevented staff from realising that the company's tax payments were up to date. In a letter of apology, a Revenue officer wrote: "I cannot give details of our Year 2000 measures and progress as you will appreciate that this is confidential, Government-sensitive information. But I can say that the above measures are causing extreme problems within the Inland Revenue, particularly this office, and have been for several weeks now."

The American government's Social Security Administration sent out more than 32,000 letters to benefit claimants with Jan 1, 1900 as the date when some of their benefits stop.

In August, the National Air Traffic Services had to switch off its computer systems for a weekend and revert to pen and paper during a year 2000 upgrade. The upgrade came shortly after NATS declared itself ready for the Millennium. NATS claimed it was for normal maintenance work.

Also in August, 4,000 Londoners were left without power over a weekend after a Millennium bug upgrade went wrong. London Electricity was attempting to make its power-key meters Millennium compliant, but when customers used the new keys, the meters were corrupted and cut off their power.

In July, it was discovered that serious faults in the Coastguard Service's new computer system could put lives at risk during the solar eclipse. The system was ordered to avoid year 2000 problems and was being installed at the time but one coastguard officer said: "It has so many glitches and faults and crashes all the time."

In June, four million gallons of raw sewage flooded a suburb of Los Angeles when a computer mistakenly closed a gate, blocking a major sewage line. The sewage overflowed 500ft on to the streets and into Lake Balboa Park. The spill occurred when city officials were testing a Y2K contingency plan.

-- Old Git (anon@spamproblems.com), December 28, 1999

Answers

Maybe it's just me....

But if I were working in banking (an industry that has gone to 'assure' the commonfolk that they are "Y2k - OK!"), and if I were in charge of sending out CD renewal notices, I would have damn well made sure that the dates were, in fact, correct. How easy would that have been? How sloppy was it not to do so? How stupid was it not to?

How telling is this?

-- Wilferd (WilferdW@aol.com), December 28, 1999.

One of my favorite examples of incompetence is the Post Office printing gazillions of stamps with Grand Canyon, Colorado on it. How difficult can it be to proofread a stamp? No telling how much that blunder cost but its all rolled into the price of a stamp. Our next price increase should be just around the corner probably $.35 now.

-- Guy Daley (guydaley@bwn.net), December 28, 1999.

Link? -s-

-- Scott Johnson (scojo@yahoo.com), December 28, 1999.

Electronic Telegraph, Scott. You wil have to take out a subscription (free). This link should point you in the right direction.

http://www.telegraph.co.uk:80/et?ac=000154642417163&rtmo=qsJsqxd9&atmo =qsJsqxd9&pg=/et/home.html

-- Old Git (anon@spamproblems.com), December 28, 1999.

As Dale Way put it in his paper, there are three areas of concern with Y2K remediation:

1. Look ahead beyond rollover
2. Rollover
3. Look back before rollover

The above article covers #1 which can be tested (and not all systems had to look ahead at once so lessons-learned could be applied to the later remediation efforts) and we are still having problems.
#2 can only be tested at rollover - either things works or they don't. We'll know in approx. 87 hours.
#3 can't be tested (in a systematic or end-to-end manner) until after rollover and all systems that must look back will be looking back at the same time with no lead time (like we have had for #1) to test the fixes.
Prognosis: since we are already having problems with #1 which we've been able to test, #3 is going to be the real test for all of the remediation efforts that have gone on. I must assume that problems will surface based on all available data.
We are already experiencing BITR. Anyone who says nothing will happen is already wrong! It can only get worse.

NOTE: Dale Way pronounced himself to be an adherent of the bell curve. What he neglected to say was what the shape of the bell curve was. Is it a skewed curve? How many standard deviations from the norm is BITR or TEOTWAWKI? I put the norm to be at 6.

-- just wondering (what.it.is@about.com), December 28, 1999.

"ONE of the fallacies surrounding the Millennium bug is that it will strike precisely at midnight on Dec 31, but it has already caused mishaps, confusion and failures several hundreds of times around the world."

Well, if hundreds of failures have happened already, and I have not been affected by them, maybe the future mishaps will not affect me either. I don't doubt that systems will fail, but I no longer believe it will even inconvenience me anymore.

-- Shawn (shawnagee@hotmail.com), December 28, 1999.

Well, Shawn, the computers don't care what you think; they're going to do what they've been programmed to do and if they have been programmed with an uncorrected date, they're likely to jam up. Since our entire way of life and physical survival is dependent upon them, you had better be right in your optimism or you may end up dead next year.

-- cody (cody@y2ksurvive.com), December 28, 1999.

Shawn,
Your argument is as inane as someone in Japan thinking that the big earthquake in the Phillipeans didn't hit them so it's no big deal...

until the tsunami sweeps up out of the sea and scours the earth bare where that person lived.

-- just wondering (what.it.is@about.com), December 28, 1999.

Shaun, There is a thread today about giving out early benefits to welfare recipients in N.J. This cost the govt. $5mil before they stopped it. The welfare recipients gave ntohing back nor were asked for anything back, the govt will undoubtedly raise taxes if more of this sort of thing happens (and it will). The Duetsche Bank is the most chilling problem reported and admitted to date. They installed a patch that caused them not to be able to close the books on Dec. 2. This means they did not clear all of their checks which affects liquidity in the entire banking system.

The Duetsche Bank problem is one where they said they could not use the backup system because it had the same patch. How could they be so stupid one wonders as to put an untested patch on both systems?

Answer: They did test that patch or it wouldn't have been put on the production systems. Here is my guess about how it got on the production systems and caused the problem. They have applications a-zz running on their production mainframes and probably only a-z in their test environment. It was application h that failed in test and they realized that while the results wouldn't show up before roll-over, they needed the patch for afterwards. After successfully installing and verifying the patch in the test environment, they rolled it out to production. Once in production, application QQ which it turns out is used in the settlement process, barfed all over itself because of the patch. This kept the settlement process from completing and necessitated the removal of the patch from both production systems. Now they have another problem, application h....... How are they going to fix it?

One other side effect here is that Duetsche Bank fell behind by a day on their transaction settlements. It doesn't take much of this before they get so far behind that they cannot catch up in a month.

You are kidding yourself in your ignorance Shaun...... It doesn't take many problems like this to disable the banking system. It is clear just from remediated applications running in current time that there are problems with dates still or you wouldn't be seeing this 1900 stuff that we are. Those are the benign problems. The ones post 1900 errrr typo there I meant post 2000 are the ones you should be most worried about.

-- WillyBoy (willy@home.in.bed), December 28, 1999.