I'm actually annoyed we still don't know the exact death rate of the virus.

I am not sure if it would be that unethical to call up volunteers that reflect the general population to establish the real death rate. (even less so if you reward them economically)

If it is orders of magnitude less deadly then that should have enormous policy implications.

Diseases unfortunately don't have a real death rate; the death rate of diseases tends to vary widely between populations for inscrutable reasons. Even if we had a study determining an 0.3% death rate among a statistically representative sample, we couldn't make policy under the assumption it's below 0.5% and couldn't be absolutely certain it's not 1%.

Yes, we could and should make policy under the best data available even if we'd prefer better ones.

I don't think that's true. It's very dangerous to make epidemic policy without accounting for model uncertainty, because by the time your model's clearly and unambiguously falsified it may be too late to deal with it.

No, not at all. Of course the model may underestimate the lethality, but it may also overestimate it so it evens out.