• Johns Hopkins University CSSE dashboard: https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6
• Wikipedia: https://en.wikipedia.org/wiki/2019–20_coronavirus_outbreak_by_country_and_territory
• King County Health Department: https://www.kingcounty.gov/depts/health/communicable-diseases/disease-control/novel-coronavirus.aspx

• Johns Hopkins University CSSE Github: https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_daily_reports
• King County Health Department: https://www.kingcounty.gov/depts/health/communicable-diseases/disease-control/novel-coronavirus.aspx

• Where confirmed cases data varies, the larger number is used. It is assumed these are more recent numbers, and that over time the sources will converge.
• Reported data will be limited by several factors, including the availability of test kits and their accuracy. As well, infected people may choose to stay at home rather than seek out medical testing. See the Scenarios tab for some models of data completeness.
• Undetected cases are a very large problem in the data. A Columbia team "estimate[d] that 86% of all infections [in China] were undocumented during the period before travel restrictions were imposed on January 23, 2020. When the transmission from undocumented infection was artificially set zero, known infections were reduced by 78.8 % across China in the simulation. This means undocumented infections could have been the source of over three-fourths of the known cases. According to the researchers, these unknown, likely mild cases seem to have facilitated the rapid spread of the virus throughout China before major control measures were implemented." Columbia team article (American Association for the Advancement of Science referenced in Chemistry views)
• With the data as flawed as it is (but no better source), the argument naturally comes up that any analysis is futile. There is merit in this argument: we are flying largely blind. But fly we must. "Catch-up" testing will ratchet up numbers for a short while (such as New York), and that will make things look worse than they are. The catch-22 is that we're damned if we do analysis, and damned if we don't. Damned we are. What we're doing is analysis on best data available, which happens to be the only data. If the data is wrong, that's out of our hands. What's in our hands is analysis, so we proceed. This is the lesser of the two evils, we believe.

• "He’s not looking at cumulative cases, but the number of new cases every day -- and the percentage growth in that number from one day to the next ."
• Article about Levitt's work (LA Times quoted in Seattle Times)
• Social distancing is critical.

• The change in the change of new cases.
• This is the second derivative of new case growth.
• Below 0 means decelerating new cases; Above 0 means accelerating new cases

Days until all infected in area (added on March 19, 2020 for US and Europe; other areas to follow) is based on 1] Population of area, 2] Existing cases, and 3] Growth of new cases in the area (see above). Our function is: LOG(Population/Cases, 1+Rate) where (1+Rate) is the base.

This is a theoretical point: densely populated urban areas are likely driving current numbers, and sparsely populated rural areas may be far less impacted by transmission. Other considerations:

• Critical assumption 1: growth in new cases based on latest recorded change, and does not take into account the direction this rate is trending.
• Critical assumption 2: the growth in new cases is solely based on recorded (historical) data, and does not take into account possible outcomes from changes in behavior such as social distancing, closures of businesses, etc..
• Durability of the virus: 3 weeks per person? (1 week hidden symptoms; 2 weeks symptoms)
  • I may infect everyone within my proximity
  • As long as I don’t move, I will not infect anyone else
• Group borders and epidemiology:
  • People inside a ring of cases are likely to be infected as the virus converges from all sides.
  • Once infected, people inside the ring are no longer dangerous, unless re-infection is possible (is it? There is a case of re-infection, but not sure what that proves.)
  • People at the edge of the ring are in danger
  • People outside the ring are not in immediate danger until the border moves toward them
  • So, the virus can only spread at the border (travelers, however, cross borders, so may destroy this containment)
  • If re-infection is possible, the distribution of the virus may become so random that there may not be any borders.