It is useful to use polarization of HF signals, but not for the same reasons you'd do that with a VHF signal, and the effects are radically different for groundwave vs. skywave propagation.
Groundwave propagation is generally considered the same as line of sight, but this isn't strictly true, as line of sight is free space propagation, where groundwave propagation is ... near the ground. As such, conductivity of the ground affects it and some refraction occurs towards the ground, extending the range of propagation on a curved surface. Horizontally polarized signals are refracted differently than vertically polarized signals, allowing vertically polarized signals to travel slightly further. (Also, the amount of refraction is larger for longer wavelengths, and different polarizations have to deal with different types of noise.)
Skywave propagation occurs when a signal bounces (really, refracts) from the ionosphere. Part of the refraction process can cause the linear polarized signal (both H and V) to be split into two circular polarized signals, clockwise and counterclockwise (faraday rotation). These signals may have variations in their propagation such that when received, they may be somewhat out of phase with each other. A linear polarized receive antenna will mix the two signals. Under certain conditions, after mixing, constructive and destructive interference effects will cause the mixed signal to fade and become stronger in a periodic fashion (selective fading). A circular polarized antenna can be used to try to receive only one of the two signals and reduce selective fading.
So, for a groundwave signal (hundreds of miles), picking your polarization may help.
For skywave propagation (more than 400 miles), the transmitted polarization will be randomized (as the path length changes), but a circular polarized receive antenna may help. In this case, transmitted polarization can't be controlled, but received polarization can be taken advantage of.
The understanding of polarization effects on groundwave propagation has been well understood for a long time, but complete understanding of the causes and effects of polarization on skywave propagation by the amateur radio community is relatively recent, and a lot of literature predates this and just says you can't polarize skywave, which while mostly true, is not the whole story. (Constructing an HF circular polarized antenna is left as an exercise for the reader.)
Edit: Here is a 1967 paper that goes into detail on this topic.
Page 6-7 describes how the signal is split into LCP and RCP signals.