Variance tensorization

For independent coordinates $X_1,\dots ,X_n$ and square-integrable $f$,For Bernoulli coordinates, each conditional variance is an explicit squared coordinate difference.