Square-1 Yoyleberry Method

[TyeDye]

I found this method just searching around the wiki a few days ago and I thought that it was very interesting. It was created by Cary Huang (Cube Roll on YouTube) and he made a video explanation of it as well here:

It is pretty much based on the fact that when a square-1 is not in cubeshape, you can control parity easier based on the fact that parity occurs when a square-1 is brought out of cubeshape.

Steps:

1.) Cubeshape - Yeah, I know I just said that it doesn't use cubeshape but it only needs it for a second. Honestly, you might be able to bypass it by grouping the edges a certain way but for now, cubeshape is the way to go.​

2.) Edge Orientation - Just worrying about the edges, get all of your top and bottom colors on their respective faces together.​

3.) Barrel/Barrel - Go back to barrel/barrel shape w/ either /(3,3)/ or /(-3,-3)/ From here, your edges should be grouped in four pairs with opposite colors together in the same orientation on all pairs (yellow edge on left, white edge on right). From now on you have to think of your edge pair as one piece. Do not split any of the edge pairs up as this will mess it up.​

4.) Quadruplets - This is the meat of the solve. You are going to solve what is called a "quadruplet" of pieces. It is two corners and two edges (one edge pair). It should look like half of a barrel shape split symmetrically.​

a.) Solve two quadruplets intuitively - Without breaking up your edge pairs, you want to match the corners up to both of the edges on one of your edge pairs. This will create a corner-edge pair-corner "quadruplet". This is generally done intuitively. Same way that you started out with square-1 and when you were learning cubeshape, they told you to just mess around with it until you got familiar with it, that's pretty much what you're gonna have to do here. Once you have one quadruplet, you can put it on the bottom left and create another on the top using a second edge pair. The way that I handle this is by treating the bottom as bandaged where I can't move the first solved quadruplet on the bottom left and only using slice moves and the U layer, I pair the corners. This also takes some practice but after a while is very easy and fast. Once you have your second quadruplet, place it on the bottom layer across from your other quadruplet. Your bottom layer should be in a barrel shape.
b.) Solve the third and fourth quadruplets - This is where the algorithms come in. In the video, he uses one swap algorithm to swap two pieces in the top left. The notation for yoyleberry is different as an edge pair is treated as one piece. With this in mind, we can use U, D, and slice moves to show the algorithms. The algorithm is this / U D' / U2' / D2' / U' / D2' / D' / D / . However, a 2-look algorithm set was proposed by Cale Schoon in the comments of the original video. http://bit.do/sq1noparity This splits the last two quadruplets up into turning the top layer into a barrel shape with similar colored corners across from each other (if it is not already) and then permutes the corners into their correct positions. The first step is fairly easy to understand and execute but the second step does have some things that make it a little confusing. With practice, it gets easier to understand the second step.
​

5 or 6.) Turn the cube back into cube shape

5 or 6.) Permute both layers of the puzzle

I perform this step before turning the cube back into cubeshape. I've recognized that when you turn a solved cube into a barrel/barrel shape, it has the top two quadruplets with the same color along the sides and the bottom two quadruplets with opposite colors along the sides. I use this information to permute the layers with the square-1 algs from Sarah's website: link. Then I turn it back into cubeshape, in which it is solved (possibly with the middle layer misaligned but that's an easy fix)
​

I really enjoy this new method of square-1 and I would like to hear thoughts from others on it as well


Edit: Forgot to add an example solve.
(4,-4) / (-3,3) / (0,6) / (-1,6) / (2,-4) / (-4,6) / (-2,-1) / (0,-4) / (0,-1) / (6,0) / (1,-1) / (-5,6) / (-2,-1) / (0,6) / (-2,-5) /

(2,-4)/(0,1)/(-1,0)/(-3,0)/ Cubeshape (Step 1)
(-5,6)/(-1,0) Edge Orientation (Step 2)
/(-3,-3)/ Barrel/Barrel (Step 3)

Since there are no already paired edges and corners, I just chose the UL edge pair to start with
(2,-1)/(-4,-4)/(-4,0)/(0,6) First Quadruplet put on bottom left (Step 4a)
/(-4,0)/(2,0)/(-2,0)/(2,0)/(4,0)/ Second Quadruplet on bottom right (Step 4a)

Now I use the 2-look algs for Step 4b (I used regular notation to keep things consistent but it's much easier on the eyes using U and D notation.)

(-4,0)/(-2,2)/(2,0)/(0,-2)/(0,2)/(0,-2)/ First Alg
(2,0)/(-2,0)/(4,0)/(-4,0)/(2,0)/(-4,0)/(2,0)/(-4,0)/(4,0)/(-2,0)/ Second Alg

I do permutation first

(-5,-3)/(3,-3)/(-3,-3)/(0,-3)/(-3,-3)/(0,-3)/(-3,-3)/ Permutation
(3,3)/(-3,-3)/ Cubeshape and finish

 

[xyzzy]

I used to use this method before I learnt Vandenbergh! The coolest part is probably that you can do 4b using 3-cycle commutators, so the whole solve requires zero algorithms. (Fun fact: when I did this, I used slice moves on both the left and the right, which looked pretty weird to a classmate who was an experienced squanner.)

I think what kills the method's potential as a speed method is that you have to go into cubeshape and then pull it back out of cubeshape to do stuff. If we could somehow combine steps 1 to 3 into a single edge pairing step (along with 1-looking step 4b), it might be pretty decent. I thought of generating algs for edge pairing, but then I got lazy and ended up learning Vandenbergh instead.

 

[Robert-Y]

Also what kills it is the inconsistent shape of the puzzle. I reckon your turning speed will reduce too much because of this.

 

[TyeDye]

I think what kills the method's potential as a speed method is that you have to go into cubeshape and then pull it back out of cubeshape to do stuff. If we could somehow combine steps 1 to 3 into a single edge pairing step (along with 1-looking step 4b), it might be pretty decent. I thought of generating algs for edge pairing, but then I got lazy and ended up learning Vandenbergh instead.

See, I think that finding new algorithms for combining the first three steps will really be the holy grail for this method, if people are only interested in it for speedsolving. I don't think that it would be very difficult, you just have to figure out how to pair the algorithms a certain way. Especially since when the square-1 is shapeshifted, it pairs edges very easily. Also, Cary has said that there should only be around 40 algs for one-looking step 4b. They really just need to be generated. I think that if we got these algs, it would have a shot at a speedsolving method.

But honestly, I look at it as the steps between beginners method and CFOP on 3x3. When you are just beginning, you have to learn 8 different steps to solve different pieces individually, sometimes messing with the same piece multiple times. But then you start to learn F2L and 2-look OLL and PLL. And you gradually make your way to full CFOP with all of the full algorithms. This is sort of this method. You have to learn the simple "beginners" algs and the step going into and then out of cubeshape and then (when they get generated) learn the more advanced one-look algs and solving edges before cubeshape.

If I knew how to generate algs, I would do it myself but I can start looking into intuitive edge pairing w/o cubeshape in the meantime. If anyone wants to help, that'd be great

Also what kills it is the inconsistent shape of the puzzle. I reckon your turning speed will reduce too much because of this.

I look at this the same way as learning cubeshape. Professional solvers know the exact moves to perform to get a puzzle into cubeshape with no pauses or lockups. I would say that it's gonna be a lot of practice but honestly, when I'm scrambling, I can do random moves with almost no pauses, probably just because I innately know what is a move and what isn't.

 

[Robert-Y]

But the pieces with cubeshape algs remain consistent every time you execute the alg, for each move. With this method, the shape of the puzzle is constantly changing. If for example I tried to do 2,0, I can probably easily do this if I have there's a corner on top to pull, but if there is a pair of edges then this isn't easy to do.

e.g. try to do 2,0 with these 2 scrambles after applying them to see the difference:
1. / 3,3 / -1,-2 /
2. / 3,3 / -1,-2 / 2,0

 

[TyeDye]

But the pieces with cubeshape algs remain consistent every time you execute the alg, for each move. With this method, the shape of the puzzle is constantly changing. If for example I tried to do 2,0, I can probably easily do this if I have there's a corner on top to pull, but if there is a pair of edges then this isn't easy to do.

e.g. try to do 2,0 with these 2 scrambles after applying them to see the difference:
1. / 3,3 / -1,-2 /
2. / 3,3 / -1,-2 / 2,0

The thing is, with almost the entire method, you stay in a situation where you only need to do moves with a multiple of 2 and it will always work. The only problem that you will have is the very beginning when you have to do cubeshape, which is the exact same as in normal cubeshape, or if we are talking about grouping edges w/o cubeshape, you'll most likely have to create a certain shape to do so, which will also use the exact same cubeshape algs, just not completed to the end. It all comes down to a familiarity with the cubeshape step in general.

 

[Cale S]

About a year ago I did a ton of experimenting with this method and generated lots of algs coming up with better ways of doing the edge grouping step. When you do edge grouping in 1 alg (not necessarily optimal since I fix the solved position of the edge pieces) and do barrel-barrel + CP in one step, the solutions are really nice and efficient

I also think it's a nice almost purely intuitive method

 

[TyeDye]

About a year ago I did a ton of experimenting with this method and generated lots of algs coming up with better ways of doing the edge grouping step. When you do edge grouping in 1 alg (not necessarily optimal since I fix the solved position of the edge pieces) and do barrel-barrel + CP in one step, the solutions are really nice and efficient

I also think it's a nice almost purely intuitive method

Hey Cale, do you think that you could help me out with that? Maybe give me a walkthrough to see what I'm supposed to be looking for? Also, could you explain how you generated the algs for the 2 look so that I could maybe generate some for 1 look as well. I'd really apprieciate it.

Btw I hope you didn't mind me mentioning your name in the original post. I was just giving credit from youtube.

 

[Cale S]

Hey Cale, do you think that you could help me out with that? Maybe give me a walkthrough to see what I'm supposed to be looking for? Also, could you explain how you generated the algs for the 2 look so that I could maybe generate some for 1 look as well. I'd really apprieciate it.

Btw I hope you didn't mind me mentioning your name in the original post. I was just giving credit from youtube.

I don't remember much about my original method other than that I disregarded it after making a method with fewer algs and didn't involve the step of intuitively grouping the first layer which was the longest and most difficult to recognize. I'm still working on a better system for it

for algs I use ksolve

 

[TyeDye]

Thanks

 

[TyeDye]

I don't remember much about my original method other than that I disregarded it after making a method with fewer algs and didn't involve the step of intuitively grouping the first layer which was the longest and most difficult to recognize. I'm still working on a better system for it

for algs I use ksolve

Hey Cale, do you happen to remember what the step for the edge grouping consisted of at all? Like a certain shape that you had to use to group edges or if you had to do two edge pairs at once? I'm having trouble grouping a singular edge pair. I'll get three of the correct pairs in a few seconds and then I'll spend ten minutes trying to pair the last one whether it's flipped around or if it's two edges across from each other on the layer. I know that you said that you gave it up but do you happen to remember just a general thing that you had to do?

 

[bobthegiraffemonkey]

Hey all, been planning to do this for a while now: optimal edge pairing statistics. Cases listed by depth.

Spoiler: cases

1:
kites
kite/muffin
kite/paws
muffins
muffin/paws
same paws
opp paws
T/6
T/33
T/51
T/123

2:
square/square
square/barrel
square/fists
kite/scallop
barrel/fists
scallop/muffin
scallop/fists
shield/fists
muffin/fists
same fists
fist/paw
U/411
U/51
U/123
L/411
L/42
L/33
L/51
L/123

3:
square/kite
square/scallop
square/shield
square/muffin
square/paws
kite/barrel
kite/shield
kite/fists
barrel/muffin
barrel/paws
scallop/paws
shield/muffin
shield/paws
U/33
L/6
L/222
T/411
T/42

4:
I/222
star/71
star/53

Average: 1.86 (28/15)

Count of cases by depth:

 0  1  2  3  4
13 16 34 24  3

Best option for edge pairing seems to be just learning them since it's at worst 4 twists and usually no more than 3. I only looked at optimal twist counts, not how to actually solve any of the cases, someone else can do that .

 

[dboeren]

Thanks for posting this. I just got a Square-1 and have been learning beginner's method but this sounds more like the kind of method I'd prefer. Will give it a test drive when I get home from work.

 

[Marlene02]

Very interesting.

 

[Aeoluz]

or you could just learn parity like wut

 

[xyzzy]

or you could just learn parity like wut

Maybe you missed the point. This method doesn't just avoid parity. It avoids the whole EP step.

 

[Aeoluz]

Maybe you missed the point. This method doesn't just avoid parity. It avoids the whole EP step.

eh.... heh... i didnt read it. Gotta admit after some reading i think i should learn yoyleberry, because in actuality i have not learnt squan yet, so should i just learn vandenbergh or learn this?

 

[xyzzy]

eh.... heh... i didnt read it. Gotta admit after some reading i think i should learn yoyleberry, because in actuality i have not learnt squan yet, so should i just learn vandenbergh or learn this?

If you want a method that looks cool, learn this.

If you want a method that's easy to be fast with, learn Vandenbergh.

In fact, why not just learn both?

 

[I_<3_SCS]

Why is Vandenbergh easier to be fast with?

 

[xyzzy]

Why is Vandenbergh easier to be fast with?

Lack of resources. Also, recognising cases is hard when the colours never stay fixed across solves, and "adjacent" or "opposite" colours have no meaning until the last few moves of the solve. Also also, what I already said in the second post of this thread, and the reply right below it.