The lack of numeracy is going to be systemic, regardless of how you state it.
Really saying "(Foo) is 100 times less than (Bar)", is just saying this:
Bar * (1/100) = Foo
It is the cognate of "Bar is 100 times larger than Foo"
Foo * 100 = Bar
I agree that it can be dicey if you say something like "X is 95% the speed of light, and Y is 50% slower (than X)" because it relies on the implied "(than X)". It COULD be interpreted as "X is 95% the speed of light, and Y is 50% slower (than the speed of light.)"
They are very different things indeed!
But, saying something is "50% colder" is just the cognate inversion of "Hotter".
You can measure thermal energy, and thus can indeed measure "coldness". Coldness is just a negative value for a thermal figure, rather than a positive one.
(EG, one can derive a "Increase in coldness" from inversion of "Increase in temperature", using a negative number for the "increase" )
Is it conceptually convoluted for no real reason, other than lack of scientific literacy, or lack of numeracy? Absolutely.
Does doing away with it actually help matters? Not really. The fundamental problem is lack of scientific or mathematical literacy. Changing the language wont fix that.
For a functional basis by which to define "Coldness", one could start with differing starting conditions for the measurement. Instead of taking absolute zero (where there is no thermal energy at all), one could start at maximally allowed entropy before the concept becomes meaningless, (derived from the mass term of the energy added by the entropic energy, against the smallest allowed massed particle, before that energy causes the particle's energy-mass to exceed the schwartzchild metric, and become a black hole) then work DOWN, rather than working UP. (After hitting that energy metric, adding more energy only increases mass, it does not increase temperature.)