Comment by Dylan16807

I'm going to do one reply to all of yours:

> I don't think people's intuitive concept of minutes and seconds is that they vary according to the time of year or the tidal effects on the Earth.

But if you asked people to choose between "seconds vary by an imperceptible amount, less than your clocks naturally drift" and "an hour doesn't have 3600 seconds" I bet most will pick the former.

And it doesn't have to drift day by day, you can average it over a multi-year period.

It's going to be weird once the earth slows down enough that we'd need a leap second every day or two. Once you can't ignore the drift anymore, the system we chose is significantly unintuitive.

> If we didn't care about "relatability", nobody would have gone to all the trouble of trying to determine how many cesium clock oscillations there were in 1/86400 of an epoch day.

I'm not saying we didn't care about it, I'm saying we didn't put it as top priority. We went with a nice clean fixed-length second that will drift away from the Earth over time. And it wasn't/isn't that hard to determine the number of oscillations since "matching" the Earth's unstable speed has a huge margin you can land within.

> Would you also say that getting rid of leap seconds and allowing UTC to gradually drift away from the sun is making the technical side easier at the expense of relatability?

Yes.

I'm in favor of both, to be honest. But it's a tradeoff, and it's a tradeoff that weakens the layman's definition of a second.