I think the argument being made is less about lasting immunity but reevaluating the actual risks for the general population. 8 million people live in NYC and there has been 0.155 million confirmed cases so far. If the actual infection rate is 20% then that represents a 10 fold overestimation of morbidity and mortality.

A good place to keep an eye on for the short term would be Sweden. Despite the lack of lockdown their disease penetrance is still on par with the UK.

> If the actual infection rate is 20% then that represents a 10 fold overestimation of morbidity and mortality.

10 fold over what? A problem since the beginning is that many people are confusing CFR and IFR. Worse is when people compare the IFR of COVID-19 to the CFR of the flu. Regardless, the IFR for COVID-19 has been thought to be .5-1% since the beginning. If we assume the NY antibody study is mostly correct (even with the sampling errors), I believe it puts the IFR in the .5-1% range [1]. If that IFR holds it still means 1.6-3.3M deaths in the US assuming the healthcare does not get overwhelmed.

[1]

deaths/(cases x 10 fold) x 100 == IFR

21908/(288313 x 10) x 100 == ~.75%

Data pulled from https://www.worldometers.info/coronavirus/country/us/ on 4/26/2020 @ 8am EST.

> If that IFR holds it still means 1.6-3.3M deaths in the US assuming the healthcare does not get overwhelmed.

You cannot assume that 100% of people will be infected. Looking at case studies like USS Roosevelt (840 of 5000) and Diamond Princess (712 of 3,711) as the worst case prevalence because they are much higher-R environments.

So basically your IFR based fatality numbers could be divided by roughly 5.

In both case quarantines were put in place and/or people were eventually evacuated. Yes, there is a limit where not 100% of the population will be infected. Given the R0 of COVID-19, currently herd immunity is thought to be reached between 60%-80% of the population getting infected. So even if we are generous and take the low end of the IFR we get 960k - 1.28M deaths to reach herd immunity.

There is some news out that is putting the IFR closer to .3% on the low end. That is great news if it holds up. The problem is that the numbers out of NY, if flawed would bring the IFR lower than reality, and they are ~.75% IFR.

However, biased antibody studies (no self-selection criteria) that may have high rates of both false negatives and false positives do not represent anything about the current level of estimation whatsoever.

Which is why when these studies happen, the public is told to wait for it to be assessed, rather than pretending all of us are remotely qualified to judge the content and draw conclusions from it about what actual risks the general population might be facing.

I'm not sure I follow what you're responding to. Antibodies are definitive proof of a previous infection, which is what I was talking about.

> Antibodies are definitive proof of a previous infection, which is what I was talking about.

When the studies in question have a high rate of false positives, that is absolutely not the case. It may simply be a statistical anomaly, from taking the incorrect confidence interval.

Currently, from the studies taken, it looks like we have high rates of both false negatives, and false positives. Which means that the testing does not give you an accurate picture of whether a population group has previous infections or not.