Quote:
Originally Posted by 'Hamas' Jenkins
Hong Kong University medical school has kept a running tally of what they believe the R0 of the virus is there. They got it as low as 0.4. So yes, I believe, it can be done.
And the model that you are talking about was assuming 50% social isolation, not 90%. That has been fairly extensively discussed on here over the last day.
If you have a virus with an R0 of 5.7 or 2.7 or 3.5 and you implement 90% of the population in social distancing then you will get the R0 below 1. The degree below one is of course dependent upon the R0, but 90% compliance to a therapy that was assuming 50% compliance will result in a fairly substantially different output, because as soon as I get that level of compliance, I'm already transmitting the virus to fewer people than currently have it, which, by definition, will move the peak in and flatten it.
Your thesis only applies if the R0 remains above 1 but below the critical threshold of the healthcare system.
|
But what they're saying is that it the R0 literally dropped to 1 overnight.
Because those models have been trending like crap from 24 hours of being released. They've NEVER been close. If they're claiming that the fact that they were off by 2-300% within 72 hours of releasing the model because of improved social distancing, then they're saying that the 90% was already happening and the R0 was down below 1 even prior to the study being released (because these are all lagging indicators).
And even if we're NOW at at R0 below 1 (I don't believe we are because again, history says it simply never happens in practice like it does on a spreadsheet), there are still lagging indicators all over that data and it STILL wouldn't explain how wildly disparate the changes were, even on a state to state basis.
Nor does it get into the fact that, again, the models lacked internal consistency. When the imperial college put forward its best case scenario and assumed complete social isolation, it didn't bring the peak forward.
You can't be off by that much, that fast and claim it was a result of a social behavior that wasn't even being demanded on a nationwide (or generally statewide) basis at the time your model was released. It simply doesn't work. Maybe had the models tracked for a week or even 2 and then went off the rails, I'd buy that.
But they were wrong immediately. And no, there's no way to say that a wholesale behavioral change that hadn't even been adopted yet caused that. Especially when they kept 'updating' the model mid-stream and still didn't think to address that claimed multiplier? Wouldn't that have been the first and most obvious target?