[wcoenen]

> a better approximation of evaluating your survival chances is to look at the death to recovery rate (dead / (recovered + dead))

This formula is still fundamentally wrong because it assumes that the time between infection and death is the same as the time between infection and recovery. I don't see why that would be true.

[Retric]

The argument is it’s a closer approximation. Abstractly if the average death took place X days after infection and recovery took more than 2X days after infection then deaths:infection is a better metric. On the other hand if the average recovery takes less than 2X then deaths:recoveries is a better metric. And if it’s close to 2X then averaging them is a better idea.

Of course detection also lags infection so that’s another consideration. I suspect that’s the real difference as outside of China a lot of cases where discovered before symptoms developed due to blood tests.

PS: South Korea has 1,766 confirmed cases, 13 dead, and 26 recovered. Iran has 245 confirmed cases, 26 dead, and 21 recovered. Those numbers are far enough apart to suggest underlying differences in reporting, which is I a larger issue. https://en.wikipedia.org/wiki/2019–20_coronavirus_outbreak