Respectfully, dividing the number of deaths by the number of confirmed cases is not a useful method to determine mortality rates. The more any host country tests, the more confirmed cases arise compared to a relatively fixed number of deaths. A good recent example is NY, which doubled the number of confirmed cases (representing a halving of the fatality rate) due to a large testing surge. Across the international data, the countries that test more have much lower fatality rates (this is in fact a power law relationship). Germany, South Korea and the Diamond Princess - places with the most effective testing protocols - have rates of 0.2%, 0.6% and 1.0% respectively, with the Diamond Princess representing a much older and vulnerable population.
Of course even these testing regimes are missing a large number of cases, and to get an accurate measure the data we do have should be fit with regression and extrapolated out to much larger testing numbers - representing more true rates of deaths among total populations. When I did that the German numbers looked the most accurate, and may possibly be overestimated.
Furthermore, there is a modeling error introduced by using one single fatality rate across demographics. What we know of the fatality distribution is that it is exponentially weighted in older demographics, meaning an average number is an overestimate. In addition, this simplification dramatically reduces the role of "herd immunity" in epidemic simulations.
Finally, if we extrapolate current and peak death rates, epidemic onset and duration times, as well as steady state rates for epidemics that have stabilized (SK and China) across the international data sets we can get a sense of how real populations, demographics and geographies are responding to epidemic conditions. When I did this I got fewer than a half million deaths globally (today it is less than 20k). For perspective 250k to 600k are attributed to the flu globally each year.
Thank you for sharing your perspective and projections. I have seen these projections from other sources, and thought hard about them. I have been preferring other predictive methodologies due to limitations I expressed above.
For the record my family and I have been social isolating for weeks before it became recommended and public policy (my family works in healthcare, and we made the decision independently based on our own prudent judgement). We continue with these practices today.
That said I believe the public and political reaction has resulted in an inefficient allocation of goods, services, and attention - and that as a result the measures being taken and the manner they've been executed dramatically outweigh the risk.
For the record I do not believe that "nothing should be done". Clearly any disease without partial immunity, vaccines, and standard medical practices poses some threat to society. I believe that the projections used in common folklore about the nature of COVID has led us to severely inoptimal strategies for dealing with the threat.