> Suppose a person is infected, but he's showing no symptoms, and doesn't even realize he's infected ... hit by a transport truck It's easy to invent a scenari

> Suppose a person is infected, but he's showing no symptoms, and doesn't even realize he's infected ... hit by a transport truck It's easy to invent a scenario that wouldn't get counted correctly. Here's one that wouldn't get counted correctly with your method: "Suppose a person is hospitalized for diabetes and would have a 95% chance of surviving, except that she also has covid which reduces her survival probability to 40%." There are two or three orders of magnitude more hospitalizations for [diabetes + covid] than [massive head trauma when hit by a truck + covid]. So maybe we should use the counting method that is least inaccurate overall, and not try to optimize for rare events?