Reading 17 Practice problem 10 (Conventional Mean Variance Frontier)

In Reading 17 Question 10, the question asks if Frontier A could be the conventional mean variance efficient frontier.

The answer says no because the conventional mean variance efficient frontier is calculated using estimates of return distribution parameters. Fine, and then it follows up with "the weights in Frontier B do not reflect estimation error, so the resulting efficient frontier must be superior to the conventional efficient frontier using estimates "

I am lost with the explanation given. Can someone help ?

I guess it refers to the fact that mean variance approach is sensitive to estimate error in expected return. So efficient frontier from actual data is superior to frontier from estimates. In my opinion, this is a poorly crafted question.