[LEAPSECS] Another reason not to mess with UTC

[Rob Seaman]

On Mar 1, 2010, at 6:21 AM, Poul-Henning Kamp wrote:


> In message <99A2C4D3-3C72-4C18-A1E1-F1B63E5D9C87 at noao.edu>, Rob Seaman writes:

> 

>> GPS is being built into critical infrastructure, [...]

> 

> Ohh, I don't disagree with that being patently stupid, without a

> good backup system.


I didn't say "backup", I said "fallback".  By all means attempt to build a system that can autonomously detect jamming and spamming and route around it - the black hats will then target the backup segments of the enlarged system.


> I just don't see it as having anything to do with UTC, and I think

> that your attempt to use it as a crutch for your crusade for the

> leap-second is just spin.


It may be spin.  It is not "just" spin.  The Earth comprises an unjammable timekeeper that is tied in a simple way into our civil timekeeping system.  Removing this feature is shortsighted.


> The backup system will have to be fully automatic, and therefore

> can use any random timescale (like GPS and LORAN does) the designers

> may choose, including a UTC without leap-seconds.


Of course, LORAN is itself being decommissioned:

	http://catless.ncl.ac.uk/Risks/25.92.html#subj11

"Fully automatic backup" is an oxymoron.  Having determined that GPS is being jammed, would other black boxes be trusted?  And if you do trust the backup as being more resistant to jamming, why not use it as the primary navigational system?

Note that jamming is physics, not computer science - simply overwhelm the original faint signal from orbit with a strong signal from your speedboat.  The implications of non-detectable spamming are in the succeeding RISKS article:

	http://catless.ncl.ac.uk/Risks/25.95.html#subj2

This raises the question of what calibration procedures a ship's pilot might use to verify the proper functioning of whatever navigational system.  Without a primary reference rooted in the real world, how can a navigator trust her tools?

Rob