As requested by @ASlavitt, here are more or less comparable pneumonia mortality numbers for Florida, Georgia and Texas (cc @Noahpinion, @JamesSurowiecki)

I used weeks 4 through 19 for comparison with past years because, while there is week 20 data available on CDC FluView for this years, it's pretty incomplete https://gis.cdc.gov/grasp/fluview/mortality.html

Florida and Georgia have definitely seen big increases in pneumonia deaths, but not as big as in the (now-deleted) @ASlavitt tweet that inspired this. And most if not all of that increase appears to be counted under the CDC's Covid death totals too https://www.cdc.gov/nchs/nvss/vsrr/COVID19/index.htm

This whole thing appears to have started with somebody on Reddit attempting entirely in good faith to make a comparison between the pneumonia death toll reported on the CDC Covid site and that from past years in the CDC Underlying Causes of Death database ...

… but for reasons that I don't entirely understand (but am willing to trust @lymanstoneky on) the methodologies used in counting deaths are different and the numbers not comparable.

As someone who wasn't a heavy user of @CDCgov data until recently, I can totally understand the mistake. Turns out not every government agency presents data in as easy to use and compare a form as, say, @BLS_gov does!

So no, doesn't seem to be any hidden explosion of Covid-19 deaths in Florida. Overall mortality in the state since the last week in January is just 101% of the norm, which compares with 129% in New York (231% in NYC), 150% in NJ, 126% Michigan ... https://www.cdc.gov/nchs/nvss/vsrr/COVID19/index.htm

Oh, and look, here was @lymanstoneky making similar points in the middle of the night (our night, not his night) https://twitter.com/lymanstoneky/status/1266255303344926725?s=20

Oh, but now it's time to fact check the fact checker! These CDC "percent of expected deaths" numbers that I cited here turn out to be … problematic https://twitter.com/foxjust/status/1266360796654813187?s=20

Those percentages don't account for incomplete data from recent weeks. The CDC is reports here that Connecticut deaths from all causes are only 51% of norm. But if you download the spreadsheet it's clear the CT data is incomplete going back to early March https://www.cdc.gov/nchs/nvss/vsrr/COVID19/

A reader ran the numbers for Florida ignoring the two most recent weeks and got a death toll 108% of normal, or about 3,500 excess deaths. Florida has reported 2,364 Covid deaths. That's probably an undercount! But I suspect most of it was early on when tests were in short supply

According to the CDC, no one has died in Connecticut from any cause since May 2 (meaning the data are obviously incomplete)