[LEAPSECS] Leap smear

[Rob Seaman]

On Sep 20, 2011, at 11:02 AM, Ian Batten wrote:


> On 20 Sep 2011, at 1036, Rob Seaman wrote:

> 

>> Ask yourself if your position is really that clocks could tick at any randomly chosen rate picked out of a hat?

> 

> No-one's saying that.  If you're asking the question "could civil clocks tick at any randomly chosen rate that is within 1ppm of what astronomers say it should be" then the answer is "yes".


Ok.  So the assertion is that 1 ppm is acceptable.  Systematic errors are different than random errors (however distributed), but let's ignore that not so minor distinction.

Let's see…1 ppm is 0.0864 seconds per day.  That is a leap second (or equivalent drift) every 11.57 days.  A leap hour (presuming such is implementable) every 114 years.  Is this acceptable?  Says who?  What process should be followed?  Who should be consulted?  If the answer is easy - then it's easy to write it down.

So let's say consensus is reached on 1 ppm, or maybe 10 ppm or 0.1 ppm.  What is this tolerance measured against?  Right!  Time-of-day = mean solar time.  What is "LOD" in all those plots?  Requirements describe the problem space.  Mean solar time is a requirement.  1 ppm would be a specification against a proposed solution suitable for evaluation by a trade-off, risks, costs, schedule, sensitivity analyses, etc.

Glad to see consensus that clocks cannot actually tick at any old rate.  What can we reach consensus on next?  What process most efficiently promotes consensus?

Rob