how bad is the quality of the US case data?

it's bad.

it's so bad that it does not even predict deaths.

in general, you'd expect a death to have been caused by an infection ~24 days earlier.

but they do not even correlate. it's faint and has the wrong sign.

but perhaps that's too long and we should be looking at a shorter timeframe to account for incubation and test results.

let's try 18 days:

this is still REALLY faint. case count predicts 4% of death count. but at least the sign is correct. (it flips at 23-22 days)

what if we drop to 10 days?

this is not crazy. assume 5-6 days infection to symptoms, a couple days of symptoms to provoke getting tested, and 2 day test turnaround. but we're getting to the edge of the "reasonableness" envelope.

and we're still only predicting 20% of outcome

here's where it gets weird:

drop to 3 days, and the R2 nearly doubles.

this is FAR too short an interval for new cases to be driving deaths unless we are only testing those who are at death's door, and the deaths figures do not support that.

they support the opposite.

R2 vaults yet higher if we move to same day

unless we're just testing dead bodies (and we're not, death to case ratios are at lows) then this tells us which way causality must flow:

barring some third factor driving both, cases are driven by deaths. this is not an absurd idea.

if people respond to reports of deaths in the media by getting tested, then this causality makes sense.

the test would be to look at the cases vs deaths 2 days earlier. no way a case today can cause a death 2 days ago.

yet the correlation holds.

this gives us a very strong indication of how causality flows. it's either deaths driving testing or it's some third factor driving both.

what's interesting is that this correlation collapses in just 2 more days.

perhaps we have very short memories?

lest people find this implausible, a check on this whole idea that people read today's paper for death counts and tailor their behavior, look at EU countries that report deaths by day vs the google mobility social interaction score (i set both as % of peak for comparability)

the swiss were the same.

so, even, were the swedes (just to a much lesser extent which is why their SIS series looks so choppy. a 2% move on 8% distancing looks bigger than on 30%)

and the lockdown loving spaniards were far less rapid to respond.

this all seems to pass the smell test.

so what's the takeaway here:

mostly, it's this: news reports affect behavior. this makes it important that we make them accurate.

they have not been.

and cases in the US do not, over any plausible interval, predict future deaths in meaningful fashion https://twitter.com/boriquagato/status/1274823561635434496?s=20

this would imply that care is warranted when freaking out about case counts esp when not accounting for testing and the age and health of new cases

it WILL lead to more distancing and economic harm (and no benefit in reducing spread. the evidence lockdowns do nothing is legion)

if cases are rising among the young that's probably a good thing. you cannot magic them away. you need herd immunity. you cannot just make everyone stay inside or wear masks until it's "safe".

a vaccine is never (if at all) going to come in time. https://twitter.com/boriquagato/status/1266747365609750528?s=20

so, getting over the fear is a big part of getting over the damage.

stop acting like this is ebola

if you are under 55 and healthy this is not more dangerous than flu. if you are under 18, you're more likely to get hit by lightning than die of covid

perspective matters

even the CDC has dropped CFR estimates to 0.4%.

they presume 35% asymptomatic. that would imply 0.26% IFR.

but 35% is WAY too low.

it's probably more like 70%. that implies 0.12% IFR which looks like a flu.

and under 65 it's VERY low.

under 50, it's 0.0015%.

but hysteria sells papers and hysteria does damage.

it's up to us to stop believing it and clicking on it.

nothing else will get us our liberty back, certainly not our weather vane "leaders" who seek to pander to panic because that's what they do.

they have and are failing us.

politicians will not do this for us.

we need to do this ourselves.

we need to lead.

let's try.

what have you got to lose?