Extrapolating Early Numbers

Reference:
   https://www.nytimes.com/interactive/2020/04/27/upshot/coronavirus-deaths-new-york-city.html
  
I've so far been unwilling to join the crowd that says "COVID-19 was overblown". The numbers seen so far support plausible values(had we done nothing), close to their original predictions. Originally the government gave a potential number of deaths on the order of 2 million, then on the order of 250k, then 67k, and now(with 79k official deaths) it's back up to a predicted 140k by August.There is of course controversy over which deaths to count as COVID-19, which to count as other causes, but there's one statistic that is most useful right now: excess deaths(expected deaths for an average year vs actual deaths). This value can't separate deaths caused from the virus to deaths caused potentially by the response, and can underestimate deaths from the pandemic if we had a drop in deaths due to other causes, but if we assume our actions didn't significantly raise the death rate by accident then it gives a reasonable ball-park figure to how bad this virus is.

New York has been hardest hit in the U.S. by COVID-19. They were hit early, hard, and long. They've likely seen an overall infection rate far beyond what the rest of the U.S. has seen so far, but obviously not beyond what is possible. Let's use them as a bellwether for the rest of the U.S., with a few assumptions:

1. Our response did not itself cause significant extra deaths. There's likely some who did die due to new delays in treatments for survivable ailments, but lacking data suggesting otherwise, we should assume this isn't a large percentage of excess deaths.
2. There's nothing special about New York's COVID-19 or their population that makes COVID-19 more deadly there than elsewhere. It isn't a more dangerous mutant version. The population is close enough to average demographics and risk profiles. Comparing deaths per cases there(if we knew the "true" count), would result in similar death-rate in other states. It is true that they have a higher population density than other areas, which will impact the rate of spread, but shouldn't on its own change the death-rate.
3. There's nothing special about their handling of COVID-19 that would result in a death rate different than throughout the U.S. Their hospitals, while overworked at times, could do little else to save lives. Those who died would have died anywhere else in the U.S. There are of course the accusations that sending stable COVID-19 patients back to their nursing homes has increased deaths by exposing vulnerable populations; while this can increase the apparent death-rate in the short-term, if we assume most would eventually get infected anyway, then it changes little in the long-run.

With these assumptions and some real-world values, we can do some basic math. New York City has had over 20k "excess deaths" as of April 25th(more deaths than expected over the duration of the COVID-19 pandemic; keep in mind these values are weeks out of date and are now higher). 16k of those have been attributed to COVID-19; some of the rest may also be COVID-19 caused, or may be side-effects of the pandemic(such as reduced treatment access for non-COVID-19 ailments). In any case, NYC has 10 million people. The U.S. has 320 million people. So...(320m/10m)*20k = 640k. Were the rest of the U.S. to see the same percentage of their population infected, with the same death rate, we would have 640k extra deaths. Even if we limit this to the 16k deaths attributed to COVID-19, that's still 512k deaths. 512k deaths may not be close to the original fears of over 2 million deaths, but keep in mind this is with a successful "flatten the curve" campaign to prevent severely overloading the hospitals. Those who needed treatment largely got treatment. New York City's experience validates the scale of the original national predictions.