As the resident 'fire stats nerd....
1. The real infection rate is unknown and unknowable. Many cases are undiagnosed and unreported. My personal SWAG is somewhere around 20% actually get reported, maybe less. Two, possibly three members of my extended family (one being a physician) probably had it, and just toughed it out.
2. Until you know the infection rate, you can't calculate mortality rate. If my SWAG (above) is somewhere near correct, then published mortality rates are inflated 5X. IOW, not good news, but also not as bad as reported.
3. Based on practical experience, intricate forecasting models frequently fail. The IHME model that the government is following is intricate.
4. It's not a terrible assumption to say that the health care system captures a fairly constant percentage of cases. So the number of newly diagnosed cases is probably a decent indicator of true new cases. If you want to accept those assumptions, it's interesting to note that my state, Utah, has been on a linear up-ramp since March 13. The linear trend "broke" yesterday, with a statistically detectable improvement in the number of new cases. The curve may have turned, or maybe it was just Sunday, and numbers didn't get fully reported.
5. The Utah data give no indication of the exponential growth assumed in the IHME model.
6. Without divine revelation, we don't know the future, so we have to make the best estimate we can. All statistical models are estimates. Almost all estimates are wrong. However, some are close enough to be useful. When this episode is over, we might have a better idea of how close the estimates were. But by then, they won't any longer be useful for planning purposes.
