the reason is that this basic reproduction number is something that should be uncovered empirically, not something that should be guessed at. the modeling is making that number up, and i'm actually assigning that process to quite a bit of the observed error.
i can throw out an infection rate as a rough bound in the context of a conditional (if the infection rate is .666 then....) as that is tautologically contained within itself so long as it's well-formed. i can also calculate an infection rate from an observed death count, and an otherwise derived mortality rate, which i'm far less apprehensive about pulling out of the existing data, even as i recommend some skepticism. however, i'm not going to guess the basic reproduction number by observing the number of found cases. i don't feel there's enough variables there to do that, even as i point to the contagion level appearing to be higher than appears to be the accepted guess, due to the nature of the curve.
for that reason, i've kept my models very simple and preferred to talk in english about the disease being "very contagious" and "not very deadly". if we could find that r-not empirically, though, i might be more willing to talk about things as functions of each other, and look at some more complex modelling. what i've been waiting for is antibody testing.
you'll excuse the pun, but this is a very crude way to measure that number that i hadn't thought of. it's inherently better than a guess, though. and, it's aligning with my general analysis that we might be burning out due to immunity, rather than "flattening the curve".