I really don't agree with this author's thesis. To me we can't make strong claims about the differences of the vaccines, and functionally we should treat all three of them as effective vaccines.
She correctly points out that experts are citing low-powered secondary endpoints in the vaccine trials to make strong statements on vaccine efficacy against hospitalizations and deaths, but then she commits similar errors by using these same limited and incompatible data points to argue substantive efficacy differences between the vaccines.
I think from a messaging standpoint it doesn't help anyone to act like one set of the vaccines is better than the other. Frankly we don't have that certainty at the moment. The trial confidence intervals are wide and overlapping. The secondary endpoints are underpowered. The endpoints and protocols are not necessarily cross-comparable. The timing and variant populations are different. Outside of the double blinded RCTs, the real world efficacy data is lower quality retrospective observational data that has its own issues and limitations as evidenced by the variable results cited in the piece.
This dovetails with why I dislike a lot of the certainty in the first doses first camp. We should be mindful of the confidence intervals, the power of the study data, and the limitations of the inferences when looking at the one dose data.
But clearly the vaccines work, clearly they are a highly effective intervention. So we should act like any vaccine you are offered is a vaccine to take that will significantly reduce virus risk.