I have the scramble if anyone wants to try it. Non lucky, 7 centers, 2 edges, and a permuted but twisted corner correct. About 4 minutes memo. Wow, beat my old PB by like 1:30!!! 
Omg, I can't believe it: 7:58.85 4x4 blind!!!!!!
b2 F L F' l2 r' R2 B f' D2 d2 b2 F' L B f2 R' B F L2 u' B' l2 b2 f' l2 R d' U F2 d' l r2 F L2 D d' U2 l2 B'
2 cube 2x2 multi is just a little less than a 3x3! Try 3 2x2x2s
)
First 4x4x4 BLD solve, after 8 frickin' DNFs, finally 18:05.46. I think memo was either 6:xx minutes or 8:xx minutes, because I just took a glance and don't really remember now.
Self-figured commutator for centers (so I guess that's one type only?) How many types are there anyway? Can anyone provide any threads? Thanks.
r2 for edges. I know commutators are much much much faster but I prefer minimal thinking during the solve, gives me better accuracy, I think.
3OP for corners. Should I switch to old Pochmann for less memorization? Suggestions and advice needed here
Anybody else use the same system? I memorized centers, corners, edges. And executed centers, corners, edges. I know this isn't the most popular sequence around because I've never seen anybody do this (most people memorize ABC then execute CBA), so any suggestions from more experienced people around here?
I can understand that, but if you restrict yourself to just a few of the most useful corner algorithms, you're safe. I just use setup moves with A perms, the "Q perm" ((RB'R'B)*3), and a Q variant (U2 R2 U2 (RB'R'B)*3 U2 R2 U2).
My memorization and execution times are almost always very close to identical. It's strange, but every time I get a really slow memo time, my execution time is also pretty close to equally slow. I'm very surprised how closely it tracks. A typical solve is just a little over 5 minutes memorization, and just a little over 5 minutes execution. I'd say equal for each phase is a good amount to shoot for.
I think Chris Hardwick runs about the same - equal ratio execution to memorization. I've seen him take a lot longer on memo than on execution on a few solves if they're really bad to memorize, though; I think he's more consistent with his execution time than I am. He's pretty much always really fast with execution, but I'm not.
I don't know why; I just got used to doing it this way and never changed. And there's really not that much connection between deciding how to maximize solved centers and actually memorizing the centers, for me at least. Chris Hardwick had originally suggested the idea of memorizing and executing in opposite order, and I tried it and had success with it, so I stuck with it. It's better to get the centers solved first so you don't have to worry about center-safeness on parity fixes, so my way works okay, I guess.
Actually, I don't think it's MUCH slower, but I will agree that if you're perfect with commutators, it seems like it should be a little slower. The moves for r2 are really quick to execute, typically. And r2 solves 1 piece with 7 moves typically, but there are cancellations fairly often (for instance, if you're doing 2 pieces and the second one is directly interchangeable with the first with an R or L face turn, you can cancel out 3 moves - so 11 moves for the 2 pieces, and it's really easy to see the cancellation as you're going). And the most important thing is that r2 requires almost zero thinking, whereas the commutators still require me to think. I'm sure Chris has gotten beyond that, but not many people have; it's a lot of work to get where commutators come to you without even thinking. I of course never have the 20 move bad cases; I use 8 move commutators to solve 2 pieces when I hit those cases. I think it's the best of both worlds.
.
