I don't see how that would do much good. Solving cross first would still make a massive number of cases to try to solve the LL, and then the F2L would still have a ton.
Perhaps Cross+F2L Edges only > Corners > ELL? Or even Edges > Corners? Personally, I'd like to look into Orient Edges > F2L minus final pair > Final pair+influence or solve LL Corner Orientation > Drastly reduced ZBLL. With multislotting, this becomes a very, very attractive method. I think the other route to go would be an L2L4 approach, where the beginning is simple and we are constantly making the later steps easier, while solving some other bit. I have the feeling that you can influence or even solve some of the later steps while placing the First layer corners, instead of just doing them alone, like in Duncan's method. Perhaps that's a source of improvement? Couple that with an X cross and you start having some extra potential. Block methods are another possible route to explore. I don't have too many specific ideas there, except possibly changing the definition of oriented during the orientation step of Petrus, so that you're left with a 2 gen finish. Any thoughts, anyone? -Mike On 1/19/07, Gilles van den Peereboom <[EMAIL PROTECTED]> wrote:
ok so should we start ? What about a method that solves Cross, then LL, then F2L... Weird... another idea ? :D Gilles