For your reading pleasure:

http://www.cnn.com/1999/TECH/computing/12/28/y2k.beyond.jan1.ap/index.html

Don't forget to read the side-bar "A sampling of Y2K glitches that have already appeared".

NEW YORK (AP) -- Y2K computer worries won't go away this weekend, even if nothing goes wrong. Glitches are likely weeks, even months, into the new year. And a few may linger until 2001 and beyond.

The Gartner Group, a technology consulting firm, estimates only 10 percent of all Y2K failures will occur during the first two weeks of January.

Yet an Associated Press poll taken earlier this month found that only 16 percent of respondents think Y2K problems will last more than two weeks. And the number who think the problems will be confined to less than a few days has increased from 22 percent to 36 percent.

Y2K planners are aware that Jan. 1 is no magic date, but they fear a quiet weekend might leave the public with a false sense of security.

"There is too much focus on New Year's weekend," said Bruce McConnell, director of the International Y2K Cooperation Center. "If you think that the only time to worry about the Y2K bug is on Jan. 1, then you're underestimating the problem."

Besides having new problems appear later in the year, glitches that strike on Jan. 1 might go unnoticed initially, even after employees return to work and restart computers. The full effects might not be felt until smaller glitches compound and disrupt business supply chains.

Several weeks must pass, McConnell said, "to have a good idea just how big an event Y2K is."

Ron Weikers, a Philadelphia attorney specializing in Y2K litigation, warned companies not to declare victory right away. Such statements, he said, could come back to haunt them.

Still, New Year's Day weekend will be a peak period for Y2K problems, and most major companies and government agencies will be watching their systems closely. John Koskinen, President Clinton's top Y2K adviser, will preside over a $50 million crisis center built for this weekend.

If there are any problems involving embedded chips that control power plants and other major equipment, Koskinen said, they would most likely strike around Jan. 1.

Beyond that, most glitches will probably be administrative, causing inconveniences such as incorrect billing -- but no catastrophe. And they'll be more manageable because they won't hit all at once.

The government has identified three crucial time periods:

-- Dec. 31, when the rest of the world celebrates the new year;

-- Jan. 1, when the new year arrives in the United States; and

-- Jan. 3, the first business day, when systems experience peak usage.

Koskinen's group will also look for trouble on Feb. 29, because some computers might not recognize 2000 as a leap year. Even Dec. 31, 2000, could be problematic because some computers might not be expecting 366 days next year.

In an AP telephone poll of 1,010 people, taken Dec. 15-19, the most frequently mentioned concern was the power supply, mentioned by a third, followed by banking and financial services, the transportation system, phone systems and food distribution. The poll's margin of error was plus or minus 3 percentage points.

Jan. 1 is not necessarily the first time a computer will encounter 2000, and some problems already have appeared.

A few years ago, some merchants began having trouble with credit cards expiring in 2000. In early October, some federal computers needed repair because Oct. 1 starts the federal fiscal year.

And in a twist from Maine, model 2000 cars were incorrectly marked horseless carriages -- the designation that the state uses for pre-1916 vintage vehicles.

Notices with 1900, not 2000, also have come from banks, courts and at least one college.

Some problems also occurred during Y2K testing, or as Y2K fixes introduced new errors.

The National Federation of Independent Business cites a recent survey that Y2K already hit one in 20 small businesses. Most glitches were fixed quickly, the federation said.

According to the Gartner Group, 30 percent of all failures will have occurred before 2000. And problems, growing steadily each quarter, will peak early in the new year. But they won't completely disappear until after 2001.

"Systems only fail when transactions are run," said Lou Marcoccio, Gartner's research director.

For example, glitches may arise when businesses finish their first billing cycle of the new year. That could happen anytime in January for monthly billing, or later for less frequent billing.

Some computers will also have to generate monthly, quarterly and annual reports, leaving room for problems later in the year.

-- Forum Regular (Here@y2k.comx), December 28, 1999

Answers

It merely shows that polling a group of people about something that they know little or nothing about is futile...

Y2K problems will not respond to polls...they will only respond to their programming (which, in many cases, will fail).

-- Mad Monk (madmonk@hawaiian.net), December 28, 1999.

First they tell people what to think. Then they take a poll to find out what they think.

-- Earl (eshuholm@tstar.net), December 28, 1999.

Glitches, schmiches. If we have water, waste disposal, heat and food it's a BITR.

-- Lars (lars@indy.net), 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 doesn't go far enough.

#1 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 as evidenced in the above article.
#2 can only be tested at rollover - either things works or they don't. We'll know in approx. 75 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. The above article missed this scenario completely.

Prognosis: since we are already having problems with #1, which we've been able to test, #3 is going to be the real indication for all of the remediation efforts that have gone on. I must assume that problems will surface based on all available data.

Lars, 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 mean is BITR or TEOTWAWKI? I put the mean (i.e., the most likely outcome) at 6.

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