Lobsters doesn’t have an NTP tag, but if you run anything that depends on NTP, the end of this post should make you a little worried.
An ntp tag would be too partisan. A time tag would work instead.
This post is an excellent argument for abolishing leap seconds. Adding seconds regularly causes issues, subtracting them will only be worse. I’m in favor of ignoring them for the next 200 years or so, by then either civilization will have collapsed or AIs can take care of it.
Or the industries that are negatively affected by leap seconds could use a time scale that’s not affected by them, like TAI or MJD. Then it’s just a matter of treating UTC like any other civil time (which it already is, de facto), by deriving it from TAI via a table of leap seconds.
I agree with things not mattering much in practice due to civilization being done in the mid-term anyway, but c’mon, can’t people try to make an effort to get at least the basic things right? Where is the craftman’s pride?
There is no fundamental reason why software dealing with time has to be such a shitshow. And if you can’t deal with leap seconds, simply pick a time without them. Problem solved.
I worked out a while ago that leap seconds will add up to something like half an hour over a millennium. Within a time zone, the difference between noon and the sun reaching its zenith varies by up to an hour. Ignoring leap seconds for a thousand years will mean that noon is off as a result of drift by less than it is off because of accidents of geography. There is no benefit from leap seconds for anyone other than astronomers and a huge amount of pain for everyone else.
If, in two thousand years, we’ve accumulated enough drift that time zones are one over from where they’d be, it’s a lot easier to adjust everyone’s time zone by one hour than it is to do a leap hour (assuming anyone actually cares about the relationship between the time and the position of the sun in 2,000 years).
Astrophysics grad student here. I’ve worked on enough of LIGO’s timing system and enough GPS/UTC/MJD conversions for multi-messenger searches to say that leap seconds are as much of a pain for astronomers as for anybody else!
You weren’t kidding.
What is the process for deciding on a negative leap second? Is it on autopilot, or would the relevant parties know to say “this would be an entirely untested scenario for essential infrastructure, we need to slow-roll this.”
How would you do a slow roll-out of a UTC leap second..?
I know Google does time smearing, but UTC doesn’t have a mechanism for that - Google is just accepting that they’ll be up to a second off from UTC at the beginning of each new leap second.
I meant literally delay doing a negative leap second until we’re convinced it will be safe. My understanding is that while it’s convenient for some purposes to have leap seconds, there’s no risk of infrastructure breaking if we just skipped them for a decade.
These people decide whether one is needed.
Thanks, I guess I should’ve stressed that it’s less about “who” and more about “what considerations are taken into account?”
Digging around, I found a pdf and a reference to a standard, which suggests it might be rather automatic: The Role of the IERS in the Leap Second.
Is this related to the drifting of magnetic poles?
Maybe? As far as I know, the actual mechanism for the Earth’s magnetic field is not well known.