In turbulent flow the friction is inertial for the most part: the chaotic flow causes the fluid in the faster moving center to interchange randomly with the fluid near the edge. This eventually allows momentum from the center to be transferred to the walls of the pipe (i.e. friction).
If the pipe walls had "rifling" (an internal swirling pattern), the fluid would rotate on it's way down. Rotation inhibits momentum exchange between the center and edges because of the Coriolis effect: a parcel of fluid moving outward will find itself moving inward at half a rotation later.
Could rifling reduce friction this way, or would the extra friction from the increased surface area offset any gains? Is there a way to intuitively see which effect will "win"?
Edit: CuriousOne pointed out a study that shows rifling seems to help: From "Experimental Study on Heat Transfer and Fluid Flow in Vertical Rifled":
"The pressure drop and the energy consumed by using the rifled tube were also found to be less than that of the smooth tube."
I don't know if this is due to the rotation or because the rifles act as riblets. This study is for only a single rifled tube, so there is still room for optimization.