[kgwgk]

> Actually, in the frequentist paradigm you could choose to run a sequential hypothesis test which will end when you've acquired sufficient data

And, as he said, you have to make adjustments to account for these interim analysis.