Dr. Dobb's Journal May 1998
Contrary to what you'll read elsewhere in this journal, the Year 2000 crisis has nothing to do with conversion, Cobol, or computers. No, the real problem is that the Year 2000 is an election year. And, even though the ballot boxes are more than two years from being stuffed, a rogue's gallery of crooks, con-men, and congressmen is already bellying up to the public trough.
For its part, the federal government showed again it's on top of things by recently establishing a Y2K task force to assist state and local governments, the private sector, and foreign governments. But no sooner had the government's press release hit the wires than the Information Technology Association of America (ITAA) weighed in to take credit, stating in its February 4, 1998, press release that "we suggested that a commission be impaneled to increase attention to solutions...to Year 2000 challenges." ITAA president Harris Miller went on to say that "the true count down to the Year 2000 begins today."
Sorry Harris, but in the real world, the Year 2000 countdown began a long time ago. In fact, according to those in the Y2K trenches, any yet-to-start project that will take more than 1.5 years to complete won't be finished in time. Underscoring this point, consulting firms specializing in Y2K conversion aren't taking on new projects that start after July 1998. And in some cases, 15, 20, or even 30 years isn't enough time to fix what soon will be broken.
Take global positioning systems (GPS), for instance. Over the past few years, millions of GPS receivers (not to be confused with GPS ground stations or satellite broadcasters) have been sold to sportsmen, geologists, archeologists, the military, taxi cab drivers, and others who want to know where in the world they are at any particular time. As part of the algorithm that lets the GPS receiver pinpoint location, GPS distributes time information. (A time differential is required to fine-tune rough triangulations when determining exact locations.) Since GPS time data is available throughout much of the world, other applications (such as financial computers) piggyback on GPS time/date information for a variety of purposes. Every millisecond, thousands of computers take time calibrations from GPS broadcasts for calculating interest on huge short-term electronic-funds transactions.
But in the GPS signal standard, dates use 13 bits to represent a time-unit offset from a conventional epoch date consisting of two fields (epoch+offset). GPS time receivers that have been programmed to update the epoch field will experience only a minor hiccup (if any). On the other hand, receivers that have this information burned into programmable read-only memory (PROMs) will likely fail because, on or about August 22, 1999, the date value will overflow this 13-bit type as satellites broadcast a new epoch. These hardcoded epoch time subsystems will think the calendar has been reset to the epoch in 1980. In short, any system that hardcodes the GPS epoch and is sensitive to the fact that "1980" is not "1999" will fail. Luckily for the vendors (but not users), most GPS time-receiver equipment is sold under a warranty that says "this equipment is not warranted to be suitable for any particular purpose."
This is only one type of problem faced by global positioning systems. There is, of course, the now-infamous literal value "19" in date types. There are also problems with other types, including type overflow problems at various dates throughout 1999; Y2K arithmetic that implicitly assumes dates later than December 31, 1999, are impossible; and implicit module-interface date-type conversions, to name a few.
Compounding the Y2K GPS problem is that it is difficult, if not darned near impossible, to construct valid GPS test cases to see what will happen at the Year 2000. Why? Because, say GPS experts, the future (time) states of the system depend on physical values (such as orbital elements and gravitational forces) that can only be accurately determined within about three months of the Year 2000.
As outhouse luck would have it, the ensuing chaos will coincide with Year 2000 elections campaigns. In a more perfect world, errant GPS systems would deposit special-interest soft money into opponents' coffers, while routing candidates to hostile speaking engagements. Alas, the best we can hope for are candidates who advertise themselves as "Y2K compliant."
--Janathan Erickson