My thoughts on Roux vs CFOP solves and mentality. In a CFOP solve, your first step is to solve a cross. In doing this, you lock yourself into the orientation of solving the cube from the bottom up. This makes sense, as CFOP is only an optimization of LBL. However, since you do this, you force yourself to think of the current state of the cube as 4 empty F2L slots. When you solve your first pair, you don't think of the cube's state as 2x2x2 block plus two adjacent edges, but rather as an F2L-3. In solving each F2L pair, you don't look for the middle layer edge and the corner to pair with it, you find the D layer corner, and then its partner. Once you reach your top layer, you solve your top face, and finish the cube. If, by some chance, solving the upper layer corners is part of the aforementioned first LL step, you first find the orientation of the corners using the U layer stickers, and then you find the permutation using sticker patterns. Basically, the entirety of the solve is based on recognition the U and D sticker colors. Since we have been so hardwired into this up-and-down mentality, when we switch to using Roux, we use the same tactics. Barely anyone goes beyond x2 y color neutrality, meaning that your up and down layers are always a certain color or its opposite, equivalent to dual color neutrality in CFOP. Your first block is nearly always done in a square+pair fashion. After the square is made, you only ever consider one of the ways to expand it into a 1x2x3, the way that keeps your D layer color along the bottom. Your second block is usually done as DR+2 pairs. DR must match the D color from your first block. Using the U color is frowned upon, and using the F or B is unthinkable. This is partially due to the way the next step is done: CMLL is recognized using U color sticker positions. If there are two U colors, all heck breaks loose. If non matching blocks are used, even our L6E suffers. We depend on U/D colors to recognize EO, and with lack of them or with overlap, we struggle enormously. We are approaching Roux all wrong. Roux is based on the L and R faces. Solving your first block should set the standard for what colors you will be looking for. Your second block should not be done as edge/pair/pair, but rather as a continuous blockbuilding to a 1x2x3. You should not need to pay attention to your D color; NMB should be embraced as the norm. We should ideally recognize CMLL by first locating the L and R stickers and then finding sticker patterns. This way, NMCLL would require no extra effort. EO does not need to be so rigidly recognized by U and D; first, we should look to the L and R color stickers. Then, Roux would become better comprehensible and optimized.
