February 1999

Facing Y2K

CIDMIconNewsletter Laura Katajisto, Microsoft Corporation

In “Y2K: So Many Bugs…So Little Time,” (Scientific American, January 1999) Peter de Jager points out that Y2K problems have been occurring for many years. For example, a centenarian was invited to attend kindergarten in Minnesota. The Amway Corporation rejected a batch of chemicals because it mistakenly believed that an expiration date of 00 meant 1900, not 2000.

It might seem an easy task to change all two-digit dates to four digits. But it is difficult to find these instances in computer code, and even when they are found, it may be difficult to make the expansion. Over the years, companies have lost parts of the source code for some computer programs making them impossible to update; programmers never expected that their software would last long enough for this to be a problem, but software has proven to be much longer lived than hardware.

It is possible to solve the Y2K problem without changing all dates to four digits. For example it is possible to divide all the years between 00 and 99 around a pivot. Years 00 through 44 can be considered 2000 through 2044. Years 45 through 99 can be considered 1945 through 1999. However, it only works for the years between 1945 and 2044. The pivot approach seems to solve the Y2K problem until the year 2044.

When doing computations with two-digit dates, it is possible to get negative numbers when an earlier date is subtracted from a later date. For example, someone born in 1985 will be 30 years old in 2015 but in two-digit arithmetic his age would be 15 – 85 = -70. A way to solve this problem is to add a positive number to both dates before the computation. If we add 28 to both numbers in two-digit arithmetic before doing the calculation we get 43 (15 + 28) and 13 (85 + 28). Then the age becomes 43 – 13 = 30-the correct result.

What is the bottom line for the Y2K crisis? De Jager believes that in the best case about five percent of infrastructures in power, transportation, and telephone systems will fail in some way-but not necessarily at midnight on the first day of the year 2000 and not necessarily catastrophically. De Jager believes that there will be some severe disruptions at the beginning of the year 2000 that will last about a month. Other less serious problems are likely to occur throughout the year.