I'm not sure I understand your statement https://www.nejm.org/doi/full/10.1056/nejmoa2035389 showed an absurdly large effect in a controlled blinded trial which was at least part of the reason emergency approval was granted.
Yes, we did observe a large effect within the first months in an immunocompetent middle-aged population, but as we all know now, that effect doesn't last very long. The placebo group has been unblinded and vaccinated, so we wouldn't know how they fared in the long run.
> but as we all know now, that effect doesn't last very long.
What, precisely, do you think we know? The vaccines were tested against the first strain so it's not surprising that they lost effectiveness at preventing infection entirely against variants like Omicron with significantly greater immune evasion but even there we still see massive benefits against severe cases. The current performance of the mRNA vaccines against Omicron is still better than many people cautioned would considered a good result for the first iteration of a vaccine created for a new virus.
Consistent waning of efficacy against infection over time has been observed regardless of variants[1].
At the same time, we didn't see "massive benefit" against severe cases in breakthrough infections in a matched cohort study[2]. This leads me to suspect that current statistical observations do not reflect reality and may well be artifacts. Paradoxically, we're also observing increased odds of Omikron infection after (two dose) vaccination.
Furthermore, we're administering boosters even to teenagers based on good faith, not good science. Hence, there still is a need for placebo-controlled trials.
This is exactly the sort of weak observational data that I am suspicious of. That same data[1] also shows that the vaccinated are much more likely to get infected, which means either the vaccine actually enhances infectivity, or the two groups are so poorly matched that they can't be compared at all. Either way, the data is weak.
If you look at the matched cohort study I linked, vaccine efficacy against severe outcome isn't anywhere near 90%, but rather 30%-50%. If the vaccine then also fails to protect against infection, as it does with Omikron, it couldn't possibly have 95% efficacy against severe disease. What could explain this discrepancy, other than statistical shenanigans?