Yes, agreed, I just meant why we won’t see “automatic quadrature” in the same sense as “automatic differentiation”. It’s always going to rely on adaptive sampling, for exactly this reason of dimensionality.