Seasonal time changes make exactly as much sense as buildings which number their floors “…, 11, 12, 14, 15, ….”
Mathematically, logically, there is no solution to these sorts of bugs. The flaw lies in the system that’s being modeled, not the model of the flawed system.
As you note, there is no such time as 2025-03-09 02:30. What’s the poor watch to do if you set an alarm for then? Logically, nothing whatsoever — but, of course, you still want it to make a noise in the middle of the night.
At the other end, the time 2025-11-02 02:30 will happen twice. Should the watch sound the alarm twice? Or just the first time, or just the second time?
Now, imagine the potential for chaos that this holds for, for example, an airline’s schedule. Even if (as is the case) everything internal uses UTC, it still has to get published for public consumption.
That’s something that Arizona gets right: no time changes, just permanent UTC-7 “Mountain Time.” We still get splashback, but not so much … mostly it’s never really knowing if it’s San Francisco or Denver that’s the same time as us, and whether the East Coast is two or three hours ahead.
Most of the world doesn’t bother with this idiocy — basically, just the US, Canada, and Europe.
b&