ok here is an example of some reasoning from peter that, if accurately reported, i find thoroughly yet badly reasoned and ultimately anti-compelling
ok here is an example of some reasoning from peter that, if accurately reported, i find thoroughly yet badly reasoned and ultimately anti-compelling https://t.co/4SMbLNa9rP
@eigenrobot COVID infection on Wuhan COVID doubles every 5.5 days once it's big enough that you have averages across large numbers. Oh and we suspect that was actually the first variant so it would've been longer. Though law of small numbers applies so all of that is pointless.
@poiThePoi @eigenrobot Note that you can estimate about how big a population you need for the average growth rate to be pretty close to accurate: > 1000 cases or so for how overdispersed the infection rate of SARS2 is. Didn't get there until January 2020 I'd expect.
@arguablywrong @eigenrobot Yeah, but that's 10 doublings. Even on the 4 day thing, it's 40 days. And then I notice that China shut down 800 Million people *in January* because they were freaked out so bad.
@poiThePoi @eigenrobot Just in terms of generations, my sim gives me a range of 5-25 generations for a single case to multiply to 1000 cases. And the generation time itself is variable, say mean 6 days, stdev of 4.8. We're talking really wide ranges in how long it takes for an index case to grow big.
@poiThePoi @arguablywrong @eigenrobot In case you’re interested, I got Pekar et al.’s data and plotted total infections, aligned at 1000.
@nizzaneela @poiThePoi @arguablywrong @eigenrobot @nizzaneela this is great, can you confirm how far back the tail can stretch? It looks like ~10 days is ~absolute fastest 1000 cases can be reached, what’s the ~absolute slowest?
@nizzaneela @poiThePoi @eigenrobot Very interesting, thanks.