Thanks for posting this! I read some details on their site, and it sounds like they are calculating it using number of deaths as their starting point. My understanding is that it works something like this = e.g. 1st death on March 24th (I’m using Alameda County in Bay Area) as an example. So calculating backwards that means that about 100 people had it (1% death rate), about 17 days earlier (approx. days until death, I think). So therefore we can calculate that about 100 cases existed in Alameda County around Mar 6th. Then if it doubles every 3 days, calculate forward from there to get projections. That is my rough understanding based on this article too: https://coronadotimes.com/news/2020/03/18/the-sober-math-everyone-must-understand-about-the-pandemic/