If the IFR is around 1% we would expect around 1,000,000 deaths if one third of Americans have to be infected for herd immunity. So that would suggest the US is 20 percent of the way there.
Given the haphazard way of calculating these numbers I would, however, put huge error bars around some (something like ±15 percentage points at least).
I too suspect areas are achieving some limited herd immunity. I don't think the behavioral changes adopted in the US have done much. Mostly, I just think less social people aren't getting it. More social people are.
https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scena...
And the opposite is now happening in the north (again). People come inside as the weather gets colder, and respiratory infections in general get far worse.
Picking FL as an example: deaths are down only about 50% since the peak in August (7-day averaged), and the numbers are surprisingly "sticky" (in the sense that they're not going down all that quickly.) For the record, FL lost 139 people yesterday; that's nearly the capacity of a 737.
139 deaths were reported; those deaths actually occurred over the last several months. Which means we won't know how many actually died yesterday for a while, but it's likely well below 139. Jennifer Cabrera posts regular updates with dates of deaths, e.g. https://twitter.com/jhaskinscabrera/status/13139124858340884...
https://public.tableau.com/profile/peter.james.walker#!/vizh...
Sorry, I didn't mean to convey that impression. Florida has been regularly reporting daily death counts that include deaths from several months ago.
I would need to know that the older, higher numbers were not also subject to the same delays.
They were, in the other direction. If you look at the date-of-death chart in the thread I linked, you'll see that for a few weeks there were consistently over 200 actual deaths per day, while the reported 7-day average in your chart never reached 200. The delay means that the reported count will be lower than the actual count when deaths are rising, and higher than the actual count when deaths are falling.
And of course we can't be sure which of those categories we're in at any particular point in time; if deaths do start to increase again, it may not be noticeable in the reported numbers for several days. But based on the hospitalization trend I believe it's probable that the current reported numbers overstate the actual deaths.
You'd like to know if people are wearing masks at church or if family get-togethers are now outdoors.
However, I'm looking at my local health department's ZIP-code-by-ZIP-code infection map of the metro area. And it seems almost entirely correlated with poverty, not privilege. Infection spread seems mostly due to "essential" workers continuing to work. That's a problematic thing to point out, because I don't think we can or will solve for that. But it's plain as day.
I get the idea of flatten the curve but states that already got hit hard don't really have any reason to impose additional restrictions they wouldn't do anything.
It's pretty obvious to people watching that the impression was created by liberal media that just wants to bash conservatives.
A cursory check of the numbers shows that it has no basis in reality.
Top 5 states:
Louisiana, Mississippi, Florida, North Dakota, Alabama
https://www.nytimes.com/interactive/2020/05/05/us/coronaviru...
In any event TX, NJ, and FL all have similar reported COVID deaths, but CA has fewer excess deaths than TX or NJ. Some of that is probably due to e.g. reduced automotive mortalities under more stringent lockdown, but I don't think the data you posted indicats that CA is e.g. worse than the other 3 states at reporting COVID deaths (otherwise we would see a much higher excess death count in CA compared to those 3 states).
That said, at my regional college in the Deep South, case numbers have dropped noticeably since the beginning of term - instead of 30 cases every day we have now ten, and there have been remarkably few outbreaks at frathouses, all that despite no organized testing.
So you'd really need to look at per capita infection rates and control for density, not absolute numbers.
For some anecdata, in my moderate county we have a mask mandate. The surrounding less populated and much more conservative areas don't. Since the mask mandate was put in place more than half of the cases in my county's hospitals have been from the surrounding counties. Despite that fact that they have an order of magnitude fewer people, and despite that fact that we have far more people living in poverty in absolute terms than they do.
CONFIRMED cases. The total number of cases is probably 5-10x that.
I live in the Deep South, and honestly I suspect that our curves have fallen simply due to a "limited" herd immunity effect (i.e. the groups of people most likely to catch COVID have already done so in large enough numbers). I certainly haven't observed any significant change in behaviors since the July peak, yet the numbers are falling like a rock regardless.