[LEAPSECS] Focus in the debate, alternative proposal

Nero Imhard nimh at PIPE.NL
Thu Jan 6 18:58:49 EST 2011

Many on this list seem to think that the whole "ITU debate" is somehow about the merits of leap seconds, while ITU's recommendation is certainly not about the merits of any mechanism that UTC may use to follow UT, but rather about scrapping the mechanism altogether and thus fundamentally changing UTC.

In my opinion, those who have difficulties with non-uniform timescales, and don't sufficiently care about UT to come up with a proposal for an alternative synchronization mechnism, should seriously wonder if they are using the the right time scale for their purposes. If it's a matter of legal obligation, they should talk to the legislators who make them use a timescale that approximates UT. I'm sure that if the (scientific) community comes up with a time standard that is well-defined and widely available, and the industry makes a case, they (the legislators) would be willing to make a switch (and consequently have their time signal broadcasts of 'legal time' follow suit).

However, given the fact that real problems do occur and that somehow it has to be UTC (for now I'll take that for granted), I would like to constructively discuss mechanisms that UTC may use to follow UT, and that may address the problems with our current leap seconds regime.

It seems that one of the main issues with leap seconds is that they occur so infrequently that real-life tests of "time handling equipment" are too rare, and there is the very real problem that leap seconds just aren't on the radar of systems designers and manufacturers who actually should care, or worse, who feel that they can get away with ignoring leap seconds and having an error once every few years "only".

So here's a proposal for making UTC more workable in the long run: let's alternate positive and negative leap seconds! Each year should have at least two (frequency increases if needed) and adjustments are made by phase reversals (e.g. two positive leap seconds in a row). Of course the goal here is to continuously "stress test" all procedures and all equipment. The induced pain is no worse than having to adjust for DST twice a year (something we all seem to cope with, although personally I hate DST more than leap seconds). This scheme directly addresses the problem of systems having a hard time following UT by placing strong incentives on getting it right.

Somewhow I can't imagine mine is a novel idea. What do you think?


More information about the LEAPSECS mailing list