Eliezer Yudkowsky comments on The usefulness of correlations

[Eliezer Yudkowsky]

Presumably X and Y have been converted to canonical form with mean 0, sd 1.