Ooh, that's rough methodology. Assuming an entire population has already been exposed and experienced the entire course of the disease, comparing it with the current fatalities to calculate fatality rate?
Stay away from stats 101.
Really? That's the bit you're jumping on?
If you actually were talking about stats, you'd point out you can't straight multiply the 0.000625 by 1000 to get the chance that one person in that thousand would die. In reality, assuming true randomness here, there would be some cases in which multiple people died, and a higher incidence than navely expected of no one at all dying.
Anyway, no, the fact that not everyone has gotten the virus is an important point. Not everyone is
going to get the virus, and the people most likely to catch it and die already have done so. Using the total population number versus the total death number is, therefore, a more accurate way of assessing the actual risk from the virus, as it better accounts for these variables which CFR to date would completely ignore.
If you'd instead like to compare a final death tally to the total population, that would certainly be even better, but judging from your apparent lack of equal disagreement with the many "models" which have been egregiously wrong or included real future numbers only at the very low end of their estimated range, I'm not sure I'd trust whichever number you would suggest plugging in.
As an aside: Turns out fear is absolutely the mind-killer. Who would have thought?
---Currently Playing: Rune Factory 4 Special, Pillars of Eternity, Yoshi's Island