GPS, the GlobalPositioningSystem, is a wonderful beast. Atomic clocks in orbit send out time-signals synchronized so tightly that, triangulating, a receiver can figure out where on Earth it is within meters or better. GPS time signals are also sources of times used for interest rate calculations throughout the global banking system and maybe other things too.

Unfortunately, GPS only allows 10 bits to encode weeks. GPS System Time will roll over at midnight 21-22 August 1999, 132 days before the Year 2000. On 22 August 1999, unless repaired, many GPS receivers will claim that it is 6 January 1980, 23 August will become 7 January, and so on. Accuracy of navigation may also be severely affected. Although it appears that GPS broadcasts do contain sufficient data to ensure that navigation need not be affected by rollover in 1999, it is not proven that the firmware in all receivers will handle the rollovers in stride; some receivers may claim wrong locations in addition to incorrect dates.

Some manufacturers have already solved the problem, but some have not.

This is how the precise rollover date is computed: The timescale origin (time zero) of GPS System Time, 00:00:00 UTC 6 January 1980, is Julian Day 2,444,244.500. A GPS Cycle is 1,024 weeks, or 7,168 days, so the first GPS rollover will occur at Julian Day (2444244.5+7168)= 2,451,412.5, which is 00:00:00 UTC 22 August 1999 AD, which is the midnight between Saturday night the 21st of August, and Sunday morning the 22nd of August, 1999.

What could be affected? It's difficult to be complete; many NTP signals originate with GPS times, and NTP goes all over the place. Obvious things that could crap out include:

	* Air and sea navigation systems
	* Weapons targeting systems
	* Bank interest rate calculations

Like all the date-dependent bugs, remediating this is a simple matter. But being certain that all the remediations have been done, and in the same way, and that new bugs haven't been introduced, and that there's enough awareness in enough places so that domino effects don't occur, is not a simple matter. August 22 has received a lot less press than September 9 or Jan 1, and we don't know how how severely dud dates may trigger data-integrity alarms, interrupting database functions and so on, so this is another one of those dates when the paranoid among us will want to hold on tight.

''Seems like another case of UncertaintyEqualsDisaster to me.  --RonJeffries''

Iceberg off the starboard bow, cap'n. Probably nothing, but wouldn't you feel happier in a lifejacket than a deckchair? But seriously, GPS rollover should be nothing significant if people do their homework here in the US. I mean, replace the dud transceivers, FixOnFail the databases and away you go. Trouble is again the idea that this happens all at once, all over the world, so it can lead to disruptions, panics, and the occasional hub failure. Probably nothing, cap'n, ...

''GPS doesn't detect icebergs. It tells you where you are. The average user of GPS can see his house from where he checks it. The majority of the rest are on an interstate highway. The rest know which direction land is because they have it drawn on the map. Ships won't get lost, planes won't fall, and if MerrillLynchPierceFennerAndSmith lose a few seconds' interest, I don't give a rat's. And I'll bet $500 that my computer won't even grab the wrong date/time from tick.usno.navy.mil. FearIsTheMindKiller. --RonJeffries''

Odd. Won't bet dinner against any Y2K disruption, but will bet equivalent on unimportant event. Shifts burden of proof. Will you bet $500 that no computer anywhere will be affected by EOW rollover? Okay, then how about no two computers? Four? Name a binary order of magnitude and a bet might be worthwhile. --PeterMerel

''One Computer. Mine. $500. I'll bet dinner on TEOTW, just don't know how you'll collect unless I buy your food now and you promise to reimburse me. Can I trust you? BTW, good job on the Litany, you got to it before I did! --rj''

GPS Bet: Burden of proof still on the wrong foot. One computer, any, you got a bet. One computer, yours, means half of all computers must be affected by EOW rollover just to get to even odds. Sucker bet. And not significant because your computer getting a dud timestamp won't cause much trouble. OTOH I'm not really certain what's the right way to gauge trouble. 

TheEndOfTheWorld bet discussion moved to YtwokBet