>That would double the function's runtime on average, without changing >the results at all in most cases (it could only be different if the >given operator has different semantics from the equality operator used >while building the statistics list).
So I agree with Tom’s reasoning.
