Who Gets Care as Washington State Hospitals Fill Up?

My natural inclination would be to use the predicted fatality and severe outcome distributions by age, as a probability distribution to sample from. But, I assume that the first person who “dies by randomness” would result in uproar. Even if you could design a likelihood model of improved outcomes, it seems possible that the right choice would be to avoid riots/fights by simply doing preference based on age.

There are lots of hospital/physician/doctor types on HN. How do you think about the trade off between outcomes and “makes more sense to people”?