Here is the thing: Created for still positive for the coldest days in Denmark and under 100 for the warmest is optimizing for ambient outdoor temperature. And much more human readable than heating one liter of water etc. I am not saying it is perfect, but I don't see why it shouldn't continue to be used as it is currently: F for weather reports and such while using C/K for more scientific pursuits.
Yet, it's quite handy that Celsius is based on the freezing temperature of water in winter. You always now when you'll have ice on the streets, cars, etc. :)
Remembering that 32 is freezing is not taxing and gives you the same amount of information (i.e., almost none because temperature alone is a poor predictor of weather).
The freezing temperature matters far less than you think. There is "latent heat" that must be added to ice to make it melt and removed from water to make it freeze.
Just the other week we had a wintry mix of precipitation at a nice and balmy 0.5 C which resulted in accumulated snow and ice, despite never being below freezing at ground level. It even warmed up through the day slightly to 1.7 C or so but the ice and snow never completely melted as there wasn't enough heat to overcome the latent heat of vaporization needed.
And as mentioned elsewhere, it's not exactly difficult to remember "32 F is freezing", even the most dense Americans have figured that out so it's another metric solution to a problem that doesn't actually exist in America.
