According to Friston's model he suggests that number is 50-80% and that it varies for each population. His example is he believes his model suggests that Germans had a larger non-susceptible portion than the UK. So the two neighborhoods in the Bronx that were slammed with over 40% have a different population makeup than other locations that were not hit as hard so his theory still stands. BTW, the Bronx has recently been averaging about 60 cases a day with a population of 1.4M in a small geographic region. As things have slowly opened the cases have not increased over the last month or so. In fact they are going down. Would suggest not many susceptibles left.
I recall reading his stuff awhile back (and Leavitt's - we had a huge exchange on Leavitt's theories back in March in this thread and on the CE board) and being at least intrigued, being a math/probability guy. In fact, if you go back to March/April, I was taking some of their approach and the data from he Diamond Princess (where ~20% became infected) to postulate that the max infected could end up being 20%. But then as things got worse and worse in Europe and the NE US and we saw some populations with 50-80% infections by PCR testing (prisons and meatpacking plants) and then saw 20% in NYC and up to 40-45% in parts of the Bronx (33% overall in the Bronx) by antibody testing, I moved more towards thinking that maybe the 55-80% being infected would be correct.
It was also hard to have high faith in a purely DCM-based (dynamic causal modeling) probabilistic approach to epidemiology for me, without some shred of scientific evidence showing how only 20% or so might truly be affected/infected. But in the last 6 weeks or so, since that first cross-reactivity finding was published (see the link below) in mid-May, I've found it to be at least more plausible, which is why in most of my long-winded sets of calcs of potential deaths, I now put in the caveat about cross-reactivity (and cures/vaccines and virus weakening) potentially greatly reducing infections/deaths, if confirmed.
https://rutgers.forums.rivals.com/t...entions-and-more.191275/page-116#post-4563258
Friston may be confident about all this, but I'm certainly not and until we have hard evidence of actual immunity from cross-reactivity (we don't yet) I think governments still have to plan for 55-80% true infections with an IFR of 0.5-10.%, which means millions of deaths worldwide being at least plausible.
Also, I take exception with his comments about variability in populations - it's hard to think that that could be so large as mean that 20% are "susceptible" (i.e., no "natural immunity" to the virus) in the UK and 33% in the Bronx. The simpler explanation is that people in the Bronx were subject to the most transmission risk, because it's a poorer area, where people often have more public-facing jobs and have to utilized public transit vs., say, Manhattan or SI, which had much lower infection rates (16-17% via antibody testing). But we can't "know" for sure which scenario is correct.
And I think he greatly overlooks the power of masking/social distancing (which are really the same thing, i.e., measures to prevent transmission). If those didn't matter, he needs to find a way to explain why most of the East Asian countries only have 50-250 cases/1MM, while Europe, the US and many others have 2500-10,000 cases/1MM, about 40-50X more (and similarly much greater deaths per 1MM) - most experts believe that masking/distancing are a huge part of why those countries have such low transmission rates. I can't believe that Friston would chalk up the fact that Tokyo only has 0.1% of its population having antibodies vs. the Bronx having 33% with antibodies being due to differences in native immunity; I assume the same will be true in SK, Taiwan and elsewhere in East Asia, but haven't seen antibody test results from those places yet. Meaning he's wrong about distancing/masking and if he's wrong about that, maybe his modeling isn't perfect either.
https://www.japantimes.co.jp/news/2020/06/16/national/science-health/tokyo-coronavirus-antibodies/
Edit: one more point: even if only 20% of 330MM are "susceptible" at 0.5-1.0% IFR, that would still be 330-660K US deaths, which is simply unacceptable, so it doesn't change what we should be doing, IMO, which is universal masking and a much better job of tracing/isolating any positives and contacts of positives. We can do all that without having shutdowns.
