The Y2K Bug Has Company in the Form of ‘Time Dilation’
Like many computer hobbyists, Jace Crouch, a professor of history at Oakland University in Michigan, decided to test the waters of the year 2000 by turning the clocks on his computers to Dec. 31, 1999.
To his relief, nothing happened. But within a few days, one of his computers using an older Intel microprocessor began acting strangely, jumping from January 2000 to December 2000.
In later tests, friends and acquaintances from the comp.software.year-2000 Internet newsgroup reported similar random jumps in time.
What Crouch had stumbled on was an odd computer phenomenon on the fringes of the year 2000 issue that typifies the vague uncertainties about the millennium bug.
The phenomenon is known by the lofty name of “time dilation,” or the Crouch-Echlin Effect. The latter half of its name comes from the Canadian programmer, Mike Echlin, who first came up with a theory about its workings.
Unlike the basic Y2K problem, which is a simple, logical problem, the Crouch-Echlin Effect stems from the interaction between some of the most obscure and complicated components in a personal computer.
Simply stated, after Jan. 1, 2000, some personal computers will suffer from erratic timekeeping. In extreme cases, Crouch and Echlin say, it can lead to malfunctions, such as the inability of the computer to recognize connected devices.
The two discoverers believe this little gremlin largely affects pre-Pentium computers that use antiquated internal clocks. They both concede that it is a petty issue in the spectrum of Y2K problems.
After a year and a half of controversy, there is no conclusive evidence that the Crouch-Echlin Effect is an identifiable computer malady. But it has also defied all attempts to be dismissed or explained.
The uncertainty over its origins has forced some of the biggest players in the computer industry to take on the issue. Intel Corp., Digital Equipment Corp., Compaq Computer Corp. and Symantec Corp. have all done extensive testing for the effect, but have been unable to reproduce it or figure it out.
Last year Compaq had to defend itself against a claim of false advertising in Britain because it stated that its new computers were ready for 2000. The complaint alleged that Compaq could not say its computers were ready for 2000 because they could be vulnerable to the Crouch-Echlin Effect.
The British Advertising Standards Authority eventually ruled in Compaq’s favor.
Tom Becker, president of RighTime Co., a Miami-based firm that specializes in regulating PC clocks and one of the staunchest critics of the Crouch-Echlin Effect, said he has gotten panicked calls from Exxon Corp. and the Federal Reserve Board over the Crouch-Echlin Effect.
“This is a scare tactic,” he said. “They’re proposing this problem is everywhere. It’s just not possible. I’m telling you, this has wasted so much energy.”
As Daniel Leviton, software architect for Symantec’s popular Y2K tool, Norton 2000, said: “I put this in the same category as cold fusion.”
The furor over the Crouch-Echlin Effect probably would have died away long ago if not for an apparent confirmation in October from Digital Equipment, which had been bought by Compaq a few months earlier. Digital issued a statement supporting Crouch and Echlin’s findings and offering to sell its customers a time dilation diagnostic program.
The statement was recently replaced with a new message that the company was unable to reproduce the effect and would no longer offer the repair program. But the company’s earlier findings have lent a validity to Crouch and Echlin’s claims that has stuck.
Mark Slotnick, who conducted the time dilation tests for Digital and is now an independent Y2K consultant, said there is no real dispute over the fact that some older computers can start up with erratic dates. He tested nearly 100 computers and turned up two that did. He has received numerous messages from computer users reporting similar problems.
Slotnick said there are many routine reasons that a computer will turn up a wrong date, such as a bad power supply or a weak clock battery. “It does happen,” he said. “The big debate is over why.”
Since the first report of the effect in August 1997, there have been several theories to explain the Crouch-Echlin Effect, variously blaming the computer’s real-time clock, Basic Input Output System, power supply, device drivers and low batteries.
Echlin, a programmer of safety shutdown software for Atomic Energy of Canada Ltd., a government-owned company that designs nuclear power reactors, said that the situation probably is caused by “not a single problem, but a multitude of closely related problems that show similar symptoms.”
His own theory deals with the interaction among the BIOS, the computer’s internal clock and the operating system. When a computer is started, one of the first tasks is for the BIOS to fetch the correct time and date from the real-time clock, a battery-powered circuit that keeps track of the date and time even when the power is turned off.
The clock has three basic phases of timekeeping. In the first phase, which lasts almost an entire second, the clock can be safely accessed with no problems.
In the second phase, the clock warns other devices that it is about to update itself to the next second. The warning zone is 244 microseconds long, during which the time can still be safely retrieved. The BIOS usually checks to see if there is a warning. If there is, it will wait until the next second to get the time.
In the final phase, the clock updates itself to the next second and will return bad data if accessed.
Echlin believes that after 2000 it can take the BIOS a few extra microseconds since it must go through additional calculations to handle dates past 2000. If the extra time pushes the BIOS access past the 244-microsecond warning zone, it will retrieve corrupted information from the final update phase, according to Echlin’s theory.
RighTime’s Becker scoffs at the theory, saying that the BIOS has no part in converting dates for the operating system. He added that even if there were additional calculations involved, 244 microseconds is an eternity for a computer--more than enough time to do even very complex calculations.
“Everything they describe is so foreign to my knowledge,” Becker said. “I really feel like a UFO debunker in this case.”
Echlin is nonplused by the criticism, saying time will eventually tell who is right on this issue. “We just happened to find this problem,” he said. “What else is out there? The experts think they know everything about the year 2000. They don’t know everything about it.”
*
Times staff writer Ashley Dunn can be reached via e-mail at ashley.dunn@latimes.com.
