I tend to agree with u Cube; after years of trying to make new methods my general conclusion is that future methods will not involved orient->permute steps, but rather direct solve. Consider OLL->PLL vs. ZBLL or 1LL, or 2-look CMLL (orient/permute) vs. direct CMLL. That is a simplification, but hundreds of experiments (that I have done) generally point to the fact that direct solve has greater potential in the long term. But you can still get extremely fast with orient->permute methods.