[JoeAltmaier]

Evidence to date outside Asia is, zero flattening. Most places. Thus my estimate.

If you want to posit some slight flattening, then 70 day? No more.

The terrible power of geometric growth cannot be resisted for long. You need to change the growth rules entirely to beat it. No half-measures will do it.

[SideburnsOfDoom]

> Evidence to date outside Asia is, zero flattening

That's not true. "new daily deaths" are fairly flat in Italy over the last week, not doubling geometrically at all.

https://www.worldometers.info/coronavirus/country/italy/

If the lockdown is working in other countries, then a similar state will be reached 3 weeks afterwards.

Daily new cases is down in Italy since peak on 21 March - that's a leading metric compared to deaths, although it is less precise because it can be swayed by changes in testing.

[JoeAltmaier]

Well I hope its true. The curves have shown 'levelling off' before, only to return to geometric growth when testing increases.

E.g. click on a country on the left, select 'logarithmic' on the lower right and see what you think: https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.h...

[SideburnsOfDoom]

Sure, more time/data is needed to know for sure. If you factor in 1 week until symptoms, another 1-2 weeks until the worst part of the illness, then yes, the effect of lock down on death rates won't be seen until 2-3 weeks have passed.

This is the UK 2.5 weeks ago: https://twitter.com/stereophonics/status/1238944498006274048

[SideburnsOfDoom]

* Average time between infection and death from COVID-19 is 23 days. https://twitter.com/michaeljswalker/status/12454153391497052...

https://www.thelancet.com/journals/laninf/article/PIIS1473-3...

So lockdown won't be reflected in death rates until _over_ 3 weeks have passed.

This is the UK, 3 weeks ago: https://twitter.com/stevethewall/status/1244249363427328003