I second that "needs more research". As a rough sanity check, so far covid death numbers roughly match excess mortality, so we're seeing most of that iceberg (I'd love to provide a better source, but see here for some quick plots I did with the our world in data files https://hmbd.wordpress.com/202... )

Well, you could modify the script (that I conveniently provided for exactly such reasons). But there's also a version with log axes and mobility data added here: https://hmbd.wordpress.com/202... Log is better for epidemic dynamics, but worse for absolute impact (like amount of people actually hit, a factor two really matters there!). You can decide on your own, but I actually did spend several hours trying different plot variants to find one that works for me.

Yup, I did comparable calculations here: https://hmbd.wordpress.com/202... ... and arrived at ~100mn infections for the US as a more reasonable estimate (you should definitely put more trust on the paper, but I find the country comparisons helpful)

I'm reading Gelman and Hill: Data analysis using regression and multilevel / hierarchical models... actually for several months now, mostly a few pages before I go to bed. It's well written, fascinating material with nice examples. At first I was just interested in the general topic, lately I've started building my own regression models in R (very easy once you get used to it) with publicly available data. So yes, you have to be some kind of data nerd to appreciate the book... but that's what you get for asking on Slashdot.