efficiently computing the variance in massive and distributed datasets c1f9fc1a13e3