> Add this to my "C++ chose the wrong kind of polymorphism to make first-class" tally.

Is it really the "wrong kind of polymorphism" if it isn't causing any problem and it didn't prevented rolling out features such as semantic support for move constructors?