Re: Israeli study: in order to conclude that this study is *not* evidence that efficacy is reduced against variants, it seems like you have to assume that every single one of the infected people in the study would have gotten infected by *some strain* anyways on the same timeline in the absence of vaccination.
Which is a pretty wild assumption - as far as I can tell, the straightforward model for infection is "every day, for each strain, you flip a weighted coin, and if it comes up heads, you get a COVID-19 case from that strain", and vaccination affects the coin weight (differently for each strain, since the distribution of strains among vaccinated cases differs from the distribution among unvaccinated cases). So it would be quite weird if vaccination had literally zero effect on the number of recorded cases. And yet this zero effect is mathematically required in order to keep the overall efficacy the same.
The fact that the observed overall efficacy matches what was seen in the clinical trials does point towards "they would have gotten infected anyways", but it's still pretty hard to explain that with a reasonable model of the mechanics of transmission/infection.
John,
I appreciate your responses! And apologies for the errors, grammatical and otherwise!
Re: Israeli study: in order to conclude that this study is *not* evidence that efficacy is reduced against variants, it seems like you have to assume that every single one of the infected people in the study would have gotten infected by *some strain* anyways on the same timeline in the absence of vaccination.
Which is a pretty wild assumption - as far as I can tell, the straightforward model for infection is "every day, for each strain, you flip a weighted coin, and if it comes up heads, you get a COVID-19 case from that strain", and vaccination affects the coin weight (differently for each strain, since the distribution of strains among vaccinated cases differs from the distribution among unvaccinated cases). So it would be quite weird if vaccination had literally zero effect on the number of recorded cases. And yet this zero effect is mathematically required in order to keep the overall efficacy the same.
The fact that the observed overall efficacy matches what was seen in the clinical trials does point towards "they would have gotten infected anyways", but it's still pretty hard to explain that with a reasonable model of the mechanics of transmission/infection.