There is now a lot of data emerging that makes it more possible to calculate the fraction of people who get the disease who die from it. All you need is serological testing of a subpopulation, and compare that to deaths from that subpopulation, something that is very easy to do. Ideally, we would have random serological testing of a wide array of subpopulations (by age, by career, by nation, etc.), so we can get the death probability for all these groups. One can only assume that data will eventually become available, because it is so easy to do and so obviously crucial. What we do have at present is a number of nations reporting serological results used to gauge their current herd immunity, and we can easily compare that to the reported deaths from that country. The purpose is to see if there are consistent numbers emerging to gauge the death probability, which in turn informs whether or not it is advisable to shoot for herd immunity via infection, vs. via vaccination. Here is an article (https://www.cnn.com/2020/05/21/healt...ntl/index.html) that reports these serological results: Sweden, 7.3% infection, 3,871 deaths, population 10.2 million, death probability 0.5%. US, 5-15% infection, 96,000 deaths, population 330 million, death probability 0.2 - 0.6 %. So the numbers are still rough, and it's never clear if the serological tests are truly randomly sampled from the same population that the deaths come from, but a consistent picture is beginning to emerge that the overall death probability across the entire population is something like 0.4 or 0.5% on average for every person infected. Obviously it is much higher for those in high risk groups.

This is a crucial statistic, because it means that to achieve 70% herd immunity via infection, one must tolerate the death of about 0.3% of one's entire population. That immediately means that if a nation does not want 0.3% of its population to die, they must slow the infection rate until vaccines are achieved. This is if the infected fraction is global to the whole population, ideally those most vulnerable could have a lower infected fraction but then again we are finding that even young people can have bad complications and can lead to death. 0.3% of the US population is a million people, or not much less than died in the 1918 Spanish flu epidemic (which achieved herd immunity via infection). The current death toll in the US is about 10,000 per week, and much of the country is talking about opening up, so if the death toll does not go below that mark, a million deaths would take about two years. If the death toll returns to its peak rate of 20,000 per week, it would only take one year, or comparable to estimates for achieving a vaccine (also highly unknown). In other words, if the death rate is allowed to return to its prior peak, a million will likely die in the US, and there will be little point of having a vaccine once it is obtained. If the death rate is maintained at what it is now, a vaccine in a year would save half a million lives, but a half a million will have died by then. These are conclusions that require no analysis or opinion, they are just facts and arithmetic. The additional number of people that suffer reduced standard of living, yet survive, is unknown and not counted in these statistics.