There are no non-Bayesians in a COVID foxhole
By Scott Sumner
There is some level of danger that pushes even the most stubborn government bureaucracies to start acting semi-rationally. Tyler Cowen has a brilliant new post that clearly demonstrates that the US has not reached that point.
But things are much worse in the UK, due to a new variant of Covid-19 that spreads much more rapidly. As a result, the UK has switched to the “first dose first” approach, which is very likely to save lives. And even if it doesn’t, the approach can be reversed at a far smaller cost than if the alternative view is correct.
The UK is already beginning to make substantial progress in vaccinating old people, who are of course much more likely to die of Covid-19:
Prime Minister Boris Johnson said 23% of all over-80s in England have now been given a dose of a Covid-19 vaccine, meaning some of the most vulnerable patients are getting the protection they need.
Given the new variant of Covid, the British are engaged in a race against the clock.
If the new variant becomes widespread in America before the vaccine is distributed, the entire country could end up being hit as hard as places like New Jersey, meaning several hundred thousand extra (unnecessary) deaths. Let’s hope our public health authorities come to their senses before its too late.
PS. For those who like numbers, here’s the sort of decision we face:
Likelihood of one dose first being the wrong approach: Very low, say 10%. Cost of adopting it if it is the wrong approach: Relatively low, say a few thousand deaths.
Likelihood of two doses first being the wrong approach: Very high, say 90%. Cost of adopting it if it is the wrong approach: Relatively high, say tens or hundreds of thousands of deaths.
You do the math.