[sgtnoodle]

Thanks for the link. For fun I downloaded the referenced CDC dataset. One thing I noticed is that the fraction of deaths per case for the vaccinated population was higher than for the unvaccinated population until July, at which point it "snapped" to match for the rest of the dataset. It seems like a curious anomaly.

https://imgur.com/a/QGJqs0S

[dragonsky]

Where older and more venerable people being vaccinated early? They still are more likely to die if infected. Then a spread of vaccination to the general population. You would have to look at a breakdown of who was vaccinated at the time you saw the lower vaccine benefit.

[sgtnoodle]

I thought about that, but it seems like whatever changed must have happened nearly instantaneously. Looking at the plot, it's a rather pronounced step change.