[Robotbeat]

A death of a vaccine or a death of COVID-19 is still a death. We should act according to reduce the total number of deaths, taking into the probability we could be wrong (and if you say that probability can ever be zero, then I don’t believe you). A 0.01% chance we could kill more with the vaccine than without is not an excuse to not deploy it IF the alternative means more likely much higher deaths.

But luckily we don’t have a binary choice. A dead vaccine is usually much safer. We can start the vaccine to high risk groups early and watch carefully, expanding access as results come in. We’ve already started clinical human trials a month ago.

[giantDinosaur]

The 'calculated risk' you talk of is such a bad way to evaluate the risk of a vaccine making things worse that I shudder to think of the consequences should such an attitude be adopted. Please, just stop, and read about the consequences of getting a vaccine wrong.

https://www.pnas.org/content/early/2020/03/27/2005456117

[Robotbeat]

I am aware.

[giantDinosaur]

Then you must be 'aware' that we can't simply calculate a probability of whether the vaccine will be worse or not. You either take a real gamble that it won't be, or you do the sane thing and actually observe long term effects in trials. There's no possible way you can assign probabilities like you can a coin flip - each vaccine and disease is far too unique and unpredictable.

[Robotbeat]

Actually, you are taking a giant gamble with not vaccinating as well. You can, in fact, assign probabilities based on historical data. Epidemiologists do this all the time. To pretend probabilities don’t work in epidemiology means you don’t understand either probability or epidemiology.

There is no safe option. Every option will kill some people. Your goal is to minimize the expected number of deaths as best you can. That’s just the way it is. That may mean doing nothing OR it may mean finding a way to accelerate a vaccine program OR some combination of expanded early access with careful monitoring to stepped expansion of the treatment access.

[tim58]

This conversation has stuck with me for several days, and I'd like to recap my thoughts on it.

I think this highlights the difference between risk and uncertainty. Risk is something calculable. Uncertainty is either incalculable or the error bars are so wide that it is practically incalculable.

I, for example, am willing to take a calculated risk and ride a motorcycle. It's a fun, economical way of traveling. I am not willing to ride a motorcycle I am unfamiliar with without a close inspection. Maybe an unfamiliar bike is in complete operational order. Maybe it has a crack in a frame wield and the frame may snap during normal operations. The error bars for what can go wrong on an uninspected motorcycle are too wide and it introduces a level of uncertainty I am not comfortable with.

The downsides of a bad vaccine have been shown multiple times in multiple places. It's _very_ bad. I'm unwilling to accept uncertainty when the downside is so large.