Fears about computer glitches won’t end Jan. 1, 2000, or even by the end of next month. There are also concerns about problems beginning Feb. 29, 2000, which will be the first “leap day” in a turn-of-the-century year since 1600.
John Koskinen, chairman of the President’s Council on the Year 2000, acknowledged the potential for problems caused by a leap-year bug in an interview yesterday on ABC’s “This Week.”
He explained that, generally speaking, the first year of a new century is “not a leap year, unless it’s divisible by 400.”
“And it turns out enough people didn’t understand that, that they programmed year 2000 as a non-leap year,” when, in fact, it is a leap year, said Mr. Koskinen.
Although there’s been a lot less publicity about the so-called “leap year” problem than the infamous bug that could bite on Jan. 1, “everyone’s been testing” their computer systems for both, Mr. Koskinen said.
“We will monitor whatever happens on the 29th, just the way we are monitoring what happens over the weekend of Jan. 1,” he said.
Jack Gribbin, spokesman for the president’s council, told The Washington Times later: “We don’t think it [the leap-year bug] is comparable in significance to the [year-2000] rollover. It represents another noteworthy day that could pose a challenge.”
But, he said, tests to date suggest only a “small incidence” of computers that won’t be compliant for the leap day. He noted that federal agencies and many businesses and other organizations tested for this when they tested for the year-2000 bug.
“This is more of a software issue than a hardware issue,” Mr. Gribbin said. Because of that, he believes operational shutdowns would be unlikely.
But other computer experts contend that some programmers overlooked the leap-day problem when they addressed the year-2000 bug. And they say this will trigger problems for some businesses for the duration of 2000.
A report out of Tokyo earlier this month said the leap-year glitch could throw Japan’s banking system into turmoil. Experts expressed fear that settlement programs may malfunction and trigger a chain of defaults.
Computerworld magazine suggests it would be unusual if there aren’t some computer problems on Feb. 29. It points out that “virtually every leap year in the computer age has had glitches,” and it cited the following episodes that occurred in 1996:
a) A New Zealand aluminum smelter literally had a meltdown when the computers shut off. They were not programmed to handle a 366th day.
b) About 60,000 people were unable to buy Fantasy 5 lottery tickets in Arizona on Feb. 29 because the lottery-machine software did not recognize the leap day.
c) Lab-equipment software used by heart surgeons at Papwoth Hospital in Britain could not function Feb. 29, so other labs had to fill in.
d) A higher taxicab fare in New York went into effect March 1. But meters programmed by one company in Queens forgot about the leap day and wrongly charged the higher rate on Feb. 29.
Computerworld notes that the rules for determining leap years are more complicated than the “every four years” rule taught in grade school. Years divisible by 100 are not leap years, except for those divisible by 400, which are. So, 1900 was not a leap year, but 2000 will be. The next century year after 2000 that will have Feb. 29 on the calendar will be 2400.
Events in the 16th century have led to this computer-age glitch. When Pope Gregory XIII developed the Gregorian calendar, he concluded that the Earth takes slightly more than 365 days to revolve around the sun, so he added one extra day every four years to compensate.
Years later, scholars determined that Pope Gregory had overcompensated for the slight irregularity. So they decided turn-of-century leap years would occur only in years divisible by 400, or every 400 years.
Said Mr. Gribbin: “The use of [only] two digits to signify a year was standard practice” in the early days of computers, so there was good reason to fear that computers would think it was 1900, not 2000, when the new year rolls in. Major computer overhauls were required to prevent that.
But Mr. Gribbin said there’s much less reason to believe a failure to identify 2000 as a leap year was commonplace. “It’s possible some programmers made the wrong judgment,” he said.