Thread by @gro_tsen

I'm being asked to summarize my position on the decision taken by the CGPM, a few weeks ago, to get rid of leap seconds. Here are my main takes on this: 🧵⤵️ •1/14

‣ Maybe it does make sense to stop synchronizing legal time with the sun, or to find a system other than leap seconds (e.g., “smoothing” as Google proposed) to do it, BUT this should require a society-wide debate, not just a vote among metrologists. •2/14

(OTOH, it makes sense to decide to technically abolish NEGATIVE leap seconds, by waiting, if one should happen, until it cancels with the inevitable next positive leap second. First I thought the CGPM wanted to do just that — but it seems they plan to scrap it all.) •3/14

‣ The argument that leap secs cause computer problems is nuts. It's trivial to implement UTC with leap seconds in a perfectly unambiguous way, the only problem is one of standardization, and it's insane that we should prefer to disagree with the sun than fix computers. •4/14

‣ For the CGPM to claim that they're just relaxing the bound on |UT1−UTC| and pretend UTC is not being disconnected from the sun, but they'll come up with a way to resync later, is hypocritical bullshit: the difficulty of resyncing will only get worse → won't happen. •5/14

‣ What will happen (if mankind survives this long…) is that as UTC drifts apart from the sun, a correction will be incorporated in the local time zone offset: so UTC will remain stuck at TAI−37s but countries will start having minute level differences in time zones. •6/14

(So we'll have even more time zone chaos, with offsets like UTC+00:59 decided at the national level. This is not necessarily worse than leap seconds, as there is already the sw infrastructure to handle this, but it's a capitulation of the international agencies' mandate.) •7/14

‣ The claim that GNSS's are coming up with their own leap-second-less version of UTC makes NO SENSE AT ALL. They're just using TAI−T₀ for some epoch T₀ which happens to be a fairly round moment (midnight) in UTC, but really, THEY'RE JUST USING TAI. •8/14

(So “GPS time” and “Galileo time” are simply TAI counted from a certain epoch T₀, and they are no more a threat to UTC than TAI, which has existed for quite some time, ever was. This whole part of the argument makes NO SENSE.) •9/14

‣ Google's proposed method of smoothing leap seconds, admittedly, represents a more significant departure from UTC. I think the correct answer here would be to have standardized it, under a different name, for computer purposes. But I agree it's a less trivial problem. •10/14

‣ Legally speaking, I believe it is ultra vires for the CGPM to take the decision to abolish leap seconds. It is simply not under the remit they got after the abolishing of the Bureau International de l'Heure (which was the uncontested authority here). •11/14

The CGPM has authority over the units, not over timekeeping. The ITU has even less authority here. Authority over timekeeping is unclear (who controls the Gregorian calendar?) but at the very least this decision needs approval from the IAU (overseeing the IERS). •12/14

‣ At the very least, there's a serious transparency issue, because when you read the resolution that the CGPM ACTUALLY passed, it does NOT say anywhere they intend to scrap leap seconds. Deciding to abolish leap seconds without even SAYING it is totally unacceptable. •13/14

More details about my position on this blog post of mine (in French, mostly overlapping what's been said above, but not completely): madore.org/~david/weblog/… •14/14