Howso? If the conversion to Kelvin is additive (as opposed to something more complicated), and multiplying in Kelvin is fine, then how is multiplying effectively-unsigned in Celsius not fine?
For a given baseline, perhaps yes?. Kelvin has its baseline inherent, for C you can pick what you like if you're clear about it. Perhaps. (Edit: and using only positive or negative temps as pointed out above)
When I say multiply by n, what I'm really doing is applying a delta of (n - 1) times my distance to the origin.
The origin with celsius isn't at zero, so you're prevented from doing what you were pretending was multiplication (at least in the most natural way of doing it).
Obviously you can negate your offset, multiply and reapply your delta.