Zvi has repeatedly cited a paper arguing that the FDA kills more people by preventing effective treatments than it saves by preventing bad treatments. Not having followed that link personally, the results suggest that pub health statisticians are miscalibrated in expectation.
One explanation for why pub health statisticians are miscalibrated in a causing-death-by-inaction direction is that they are punished for deaths caused by action but not deaths caused by inaction. I’m this model, conservatism (aka miscalibration toward inaction) is a result of the lived and publicized experiences of people in the field. This seems a great explanation for both the experts norms and the paper results.
Are you contesting that statisticians are miscalibrated in expectation in the utility they cause?
I think the “Bailey to your Motte” is that people are bad at predicting who they might infect, so this advice could lead to greater deaths. I think Zvi could have phrased it more carefully. But the broader point needed emphasis, that we are loosing so much for something we could fix. And that fix might not be so hard. That point is more important than quibbling one darn sentence.
I am contesting the whole Extremely Online Lesswrong Way<tm> of engaging with the world whereby people post a lot and pontificate, rather than spending all day reading actual literature, or doing actual work.
Zvi has repeatedly cited a paper arguing that the FDA kills more people by preventing effective treatments than it saves by preventing bad treatments. Not having followed that link personally, the results suggest that pub health statisticians are miscalibrated in expectation.
One explanation for why pub health statisticians are miscalibrated in a causing-death-by-inaction direction is that they are punished for deaths caused by action but not deaths caused by inaction. I’m this model, conservatism (aka miscalibration toward inaction) is a result of the lived and publicized experiences of people in the field. This seems a great explanation for both the experts norms and the paper results.
Are you contesting that statisticians are miscalibrated in expectation in the utility they cause?
I think the “Bailey to your Motte” is that people are bad at predicting who they might infect, so this advice could lead to greater deaths. I think Zvi could have phrased it more carefully. But the broader point needed emphasis, that we are loosing so much for something we could fix. And that fix might not be so hard. That point is more important than quibbling one darn sentence.
A whole paper, huh.
---
I am contesting the whole Extremely Online Lesswrong Way<tm> of engaging with the world whereby people post a lot and pontificate, rather than spending all day reading actual literature, or doing actual work.
which literature do you recommend?