Just wanted to call out that I think calculating death rate as "deaths / infections" works for past infections but is wildly inaccurate today because the majority of infections have no resolution yet.

I'd say "deaths / recoveries" or maybe "deaths / (deaths + recoveries)" is probably better, with the caveat that it is likely much higher than the final number it's going to settle at as more people recover.

I've been using recovery numbers from https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.h... .