# Random Cubing Discussion

#### MaeLSTRoM

##### Member
need to make a ECE pair to finish the 3x3x4
need to pair the rest of the edges and fix EO parity and fix CP for 2gen
Whoops, forgot to put that in. I was thinking more for cubes>=5x5, I think it would be quite slow on 4x4 relative to reduction/K4/whatever...

Actually Nevermind, just gave it a go, and its much slower than reduction.

Last edited:

#### irontwig

##### Member
Generally speaking direct solving methods get worse as the cube size increases.

#### miniGOINGS

##### Member
Generally speaking direct solving methods get worse as the cube size increases.
Any known specific reason?

#### MaeLSTRoM

##### Member
Generally speaking direct solving methods get worse as the cube size increases.
That's probably true, it was just a random Idea, hence the random discussion thread.

And another thing: If on an even layered cube, you placed the corners in a solvable position, is every combination of the other peices solvable?

#### irontwig

##### Member
Yes, at least with a non-super cube.

#### Kirjava

##### Colourful
And another thing: If on an even layered cube, you placed the corners in a solvable position, is every combination of the other peices solvable?
Yep.

Anyway, I don't think every direct solving method suffers as badly.

Cage is pretty bad because of all the centre comms. I think K4 and Redux give comparable times on 7x7x7. If I could do centres and pair four edges as fast as dan our times would be about the same.

#### Kirjava

##### Colourful
Rawrby4 columns are pretty suck on biggercubes I guess.

#### qqwref

##### Member
Even I don't do this. 6x6x6 sucks to turn when it isn't on qCube.
Surely some people have good 6x6x6s, though, right? I mean, considering the tps some people get.

Would this be a good way to solve big-cubes:

3 centres around one corner
Pair the 3 edges + 1 corner to make a N-1xN-1xN-1 block
Finish centres
Pair up one CE pair so that only R&U are left to solve
Solve petrus-style 2-Gen stage
Nope. Once you've got those 3 centers made, there's no way to pair edges other than the standard reduction style (2 or more at once). So this can't be more efficient than Reduction+Petrus, and will probably be worse, because the order of steps doesn't always let you take advantage of easy cases on future centers/edges.

Roux-by-4/KBCM doesn't work so well on bigger cubes, because the centers+columns with <2R,3R,...,U> step starts to break when +centers get involved. As far as K4 on 7x7x7, no clue whether this can compete with Redux on the highest level (I'd have to try it on the sim a few times but I can't think of a nice hand position for it). It's entirely possible that Dan's edges are lacking in lookahead, or that Kir's edge comms have very high tps.

Last edited:

#### oll+phase+sync

##### Member
...such as lighting ones always being on the left
when I see a lightning bolt I always AUF, so the oriented Corner is at UFL (Sune case) or UBL (Antisune case)
The Sune/Antisune case can be predicted by looking at any corner. And my algs are that way that I don't need additional AUF afterwards.

#### Godmil

I forgot to mention that I'm also doing the same thing with PLL. a 1x3 block (Ja,Jb,F) always goes on the Left hand side (Except U-perms that are easy enough to spot and I solve from two orientations each anyway), 2x2 goes in UFL (I changed my A perms so that they fitted my V-Perm orientation, so I just place the block and know the PLL based on the stickers to either side of the block). All I need to do is redo my R-Perms (which I've never been happy with) so that they have headlights on the left and then they'll match up with my G-Perms.

#### Rpotts

##### Member
I thought that would be useful back when I was finishing PLL, but it limits your alg choice too much. Just get used to recognizing from multiple angles, after hundreds of solves you sorta just "auto-AUF" without really thinking. You should be able to recognize V perm by simply seeing a 2x2 block + diagonal corners (opp stickers on each face) or by seeing the pattern it makes on the other side of the block.

#### Godmil

oh, I'm getting better at pll recognition, but I suck at some angles (the non block faces of V-perm, or any G-perm angle where you can't see the 2 block), I think I'll need to spend time properly working on 2 face recognition.

#### TMOY

##### Member
Cage is pretty bad because of all the centre comms.
Lol, there's much more in solving centers last than just niklas niklas niklas. Learning to use 4-move comms and thus reducing the move couny by a lot allows to go much faster than you seem to think.

#### Cool Frog

##### Member
Lol, there's much more in solving centers last than just niklas niklas niklas. Learning to use 4-move comms and thus reducing the move couny by a lot allows to go much faster than you seem to think.
lol
true

Last edited:

#### DavidWoner

##### The Punchmaster
I forgot to mention that I'm also doing the same thing with PLL. a 1x3 block (Ja,Jb,F) always goes on the Left hand side (Except U-perms that are easy enough to spot and I solve from two orientations each anyway), 2x2 goes in UFL (I changed my A perms so that they fitted my V-Perm orientation, so I just place the block and know the PLL based on the stickers to either side of the block). All I need to do is redo my R-Perms (which I've never been happy with) so that they have headlights on the left and then they'll match up with my G-Perms.
This kind of stuff will hurt you in the long run and probably not help in the short run.

#### Kirjava

##### Colourful
Lol, there's much more in solving centers last than just niklas niklas niklas. Learning to use 4-move comms and thus reducing the move couny by a lot allows to go much faster than you seem to think.
My main method used to be cage, no need to patronise.

I was talking about the problems with using so many inner slice moves that alternate between adjacent axis. Not great to fingertrick.

#### ben1996123

##### Banned
So... Lightake has this:

named as "LL Color Heart Rice Dumpling Magic IQ Test Color Cube White"

I lold.

Last edited:

#### uberCuber

##### Member
rice dumpling lol

IQ test, lol