Why 1999?

by Scott Olmstead

If you have read anything about Y2K, it's that the computers may stop working on January 1, 2000. Elevators may stop running. Planes may not fly (the FAA may shut down the airports). Businesses may close.

This is not correct. Problems will hit with force in 1999. In that year many organizations will start entering "00" as the year that a project is to end or an account is to expire, or something similar.

Systems will grind to a halt, either not accepting "00" or crashing. See USA Today, For Some, Year 2000 Is Today, which said "The scattered snags could grow into wider outages a full year before the millennium - the first business day of 1999 - because so many firms work a year in advance, from projecting profits to booking halls."

And, don't forget, fiscal 2000 starts October 1, 1999 for the federal government, and July 1, 1999 for most states!

The Topeka Capital-Journal editorialized on February 7, 1998:

It may sound as if we have two years to work this out, but we don't. Computer glitches have already begun appearing, especially in private-sector computers, which must track customer accounts and the like well into the future. As we move closer to 1999, the problems will escalate. In reality, this problem should be solved not in two years, but (by December 31, 1998)!

In fact, for many organizations, the deadline is actually sometime in 1998! In mid-1998 they will begin fiscal 1999 (for exaxmple, 44 states begin fiscal 1999 on July 1, 1998). So what is so bad about fiscal 1999? Two things: Year-long forecasts, contracts, and other events as already mentioned, and something called the Year 99 Problem.

The Year 99 Problem arises because "99" was often used as a code with a special meaning, such as "no termination date" or "delete this record" or something else. Start looking for Year 99 problems when New York State begins fiscal 1999 in April 1998!

You can guess what happens a year later in 1999: organizations begin fiscal year 2000. At that time "00" will have to be entered into the computers. Then we will really begin to see failures and shutdowns.

Consider also that if entering "99" or "00" doesn't crash an organization's computers, then the huge amount of data exchanged among organizations means it may crash someone else's computer. This is the Domino Effect. Bad data will be bounced around until it does some damage somewhere.

Another reason that 1999 will be a year of crisis is that some bugs will be put into working systems. The entire federal government and many large firms have declared that Dec 31, 1998 is the last day they will be fixing code and 1999 will be the year of testing. Don't believe it. Many fixes will be rushed into service without proper testing in fiscal 2000 (mid-1999) and produce major problems.

So expect that gradually from mid-1998 into 1999, organizations that have not completed repairs will shut down parts of their operations for short or long periods as they try to cope. The consequences of Y2K will likely phase in gradually over a period of months in 1998 and 1999, not (necessarily) as a sudden shutdown on January 1, 2000.

The 1999 Problem

by Gary North

Comment:

Some major computer problems will begin in early 1999, growing worse in the fall of 1999. This gives us even less time to prepare.

This is from the MILLENNIUM JOURNAL (Nov. 1997), published by Data Dimensions, a y2k repair software company.

Link: http://www.data-dimensions.com/html/milj45.htm * * * * * * * *

The digit "9", used in the data singly or as a series of nines, since the beginning of data processing, has held special meanings. The meanings could be end of record, end of file, error return, cancellation, start, special discounts, missing date, or keep forever, to name a few.

In some cases, the conditional nine is included in the programming language itself to have special meaning. In fact, "9" is the most frequently used element to mean transfer of logic.

During one of the early project reviews (1991), a client told a story about a new financial system that had been tested post 2000 in processing, but when 1999 was used, the system only returned error code messages! (Took 90 days to fix and retest.) The problem and its immediacy are discussed in greater detail in this article - we believe this problem is very real and urgent! ...

The date 12/31/99 is the last day of 1999. When the clock strikes "12" midnight on 12/31/99, computer failure should be expected if proper care is not taken before this date to resolve the two-digit date storage problem. However, computer failure can occur before 12/31/99, if you have an earlier Event Horizon.

Remember that the term, Event Horizon, refers to the last date that applications will process dates correctly. For more information on Event Horizons, see the Millennium Journal, Volume I.I.

Event Horizons can vary between Enterprises. For example, the U.S. Government's fiscal year 2000 begins on 10/1/99. This moves the 12/31/99 Event Horizon to 10/1/99. There are many factors specific to each program that can change an Event Horizon. . . .

Background on the "99" Problem

Many programmers did not think that their software would still be in use in 1999. Therefore, programmers often coded date fields with "99". Using "99" as program logic was standard practice at the time the products were created. The value "99" frequently was used to mean the following: end of record, end of file, error return, cancellation, start, keep forever, or other special commands. . . .

Another vulnerable area exists because of standard-language date editing, initialization, and validation rules. At the time many of these programs were written, the programs often required a number to be entered in the date field, even though a date might not be applicable. To accommodate program requirements for an entry, programmers designed programs so that users frequently enter 1/1/99 or 9/9/99 in date fields that do not have a specific date. . . .

There is yet another problem with the year, 1999. Many software programs perform year-end projections. In 1999, the computer will make projections about the year 2000. When projections are made for year 2000 in 1999, there could be complications because the year-end projections will require a date field. If your date field only allows for two-digits, then 2000 is going to be "00". With the field "00", your computer may interpret "00" as 1900. It is even possible that "00" has program logic assigned to it.

The Crash of 1999

Most information about the Y2K problem lists 12/31/99 as the typical Event Horizon. Although 12/31/99 has been widely publicized as the critical date for most systems' Event Horizon, 1/1/99 or 9/9/99 may actually be your organization's Event Horizon. Regardless of what other systems do, what will your computer do when it rolls over to the next "logical" date? . . .

This systemic failure due to "the nines" is potentially very damaging. It is no major exaggeration to say that every date processing program is potentially vulnerable to this error. And, it is almost too late to fix the problem before failure occurs. Unless immediate action is taken to address and solve the 1999 date problem, computer failure is inevitable.
***************************************************************************

[RevFile Note] If all this wasn't enough, there is yet one more "fly in the ointment" that hasn't, as far we know, found much press. Take the situation discussed in this report relative to the Federal Government's switch to their new fiscal year in October. In October 1999, the switch is into the year 2000. But, if the '99 Bug is also as serious as Y2K, then what will likely happen to government computers when they switch to the 1999 fiscal year -- in October 1998?! How many government computers, programs, agencies and social services become adversely affected by 1999 shutdowns? Could that first domino begin its cascade in the Fall of this year?

And (making matters a wee bit worse still), if the Federal Reserve system required a one year testing of the entire banking system for Y2K in 1999, shouldn't they also have planned for an additional year for testing their '99 fix? That testing year should have been 1998, meaning that the last year for curative work for both the '99 Bug and the Millennium Bug should have been in 1997! It appears, from here, that we have already crossed the point of no return. [RFN] ****************************************************************************