The New Method / Substep / Concept Idea Thread

[Wish Lin]

I seriously believe @Angry_Mob ’s method can be a match with full CLL on some worse cases! Sune or antisune’s permutation effect on the last layer is definite, so the last layer can be one one-looked. Seems familiar, huh? It’s just full CLL with longer algs!

 

[xyzzy]

What about solving a flipped f2l pair and then fixing it during OLL?

OLS-FE.

Or PLL + 2flip?

Algs will probably be terrible. Also, both of these suffer from not being slot-neutral unless you want to learn four variants of the same alg (or do extra rotations, like the y2 you have in your first sample solve).

 

[Wish Lin]

Why not simply influence last layer while solving the first layer to force an H or Pi CLL case? It's hard to do every solve, but it would be much more efficient than tacking on 7-8 moves every solve.

I thought about that as well, but that will require some 2x2 version of VLS, and you also have to take the observation time into acoount, so...... Just use EG, less algs.

 

[Wish Lin]

New 2x2 thing?

This is a method for people who are learning or want to know CLL.
It's essentially using sunes to set up cases, but imo treating it as an alg gives it merit.

The Steps

1. Layer - just like CLL and LBL

2. Use a sune or anti-sune to set up a case w/ no corners oriented. These sets (Pi + H) have the easiest recognition imo, there are only 10 algs and almost all of them are good. It's good to treat each case in this step like an algorithm, even though there are only two.
Here are the angles (starting angles are from here http://www.cyotheking.com/ortega )
S - U' (AS)
AS - U2 (S)
U - U (AS)
T - U' (S)
L - U (AS or S)
Personally, I prefer backsune, but you can figure that out yourself

3. CLL - only 10 cases show up. It is impossible to skip this step which kind of sucks.

Overall I think this is pretty good because of its reasonable alg count, easy recognition and low move count (at least compared to LBL)
I don't know if this is 100% better than LBL, but it has its advantages.

With @Angry_Mob 's agreement, I come up with an almost identical idea with this: Instead of reducing to and Pi or H case, this reduce it to a Sune /Antisune case. CLL algs for those two cases are much shorter.

Method PDF:

 

[Wish Lin]

With @Angry_Mob 's agreement, I come up with an almost identical idea with this: Instead of reducing to and Pi or H case, this reduce it to a Sune /Antisune case. CLL algs for those two cases are much shorter.

Example solves(using @Angry_Mob 's SC):

LBL v.s. His method v.s. My method

1.
Scramble:
U' F U' R' U' R U2 R' U2

LBL:
27 moves

His:
25 moves

Mine:
x F R U2 R' // FL
R U R' U R U2 R'// Reduce
F' L F L' U2 L' U2 L U2 // CLL+AUF
20 Moves

2.
Scramble:
R U' F R' U F R2 F U' R'

LBL:
31 moves(although there are shorter Y- perms)

His:
23 moves

Mine:
x2
F' R' U' R2 // FL
L' U' L U' L' U2 L// Reduce
L' U2 L U2 L F' L' F U2// CLL+AUF
20 moves(or 13 moves if you cancel the solution which is just a lefty T OLL)

Hope this method can be useful!

 

[Imam Alam]

I kept adding more and more details to my previous posts in this thread (in order to avoid double or triple posts), so much so that after about 2 months those posts have now become huge, with spoilers inside spoilers (spoiler-ception lol)

so I thought I should make this post and settle this thing once and for all, for the benefit of those users who had a look earlier and missed the details added later

below are links to those said posts, and a list of new information added:

.
.

1. collection of methods
https://www.speedsolving.com/threads/the-new-method-substep-concept-idea-thread.40975/post-1315558

Spoiler

added 3 new methods (ZZ, Petrus, sandwich)
sorted methods list (again)
clarified potential confusion in algorithms list
did general tidying up
added relevant quotes from famous cubing figures

2. new method proposal (HSC)
https://www.speedsolving.com/threads/the-new-method-substep-concept-idea-thread.40975/post-1315749

Spoiler

added further details on how to implement the method (further instructions)
added fundamental techniques section
added more details on algorithms (especially CO) along with flowcharts and diagrams
added ways to optimize for various purposes
described more variants
added relevant quotes from famous cubing figures

3. new method proposal (2GB)
https://www.speedsolving.com/threads/the-new-method-substep-concept-idea-thread.40975/post-1318959

Spoiler

added basic description of the method and how to implement it (basic instructions)
added further details on how to implement the method (further instructions)
added fundamental techniques section
added example solve
added further details on algorithms (especially CO) along with flowcharts and diagrams
added ways to improve the method and make it more efficient
described ways to optimize for unique lossless variant (once a block is built, it is never broken -- not even momentarily -- until the entire cube is solved)
added relevant quotes from famous cubing figures

hopefully this would not be considered as undesirable bumping (or shameless self promotion) lol

 

[PapaSmurf]

I'm not sure if I've ever proposed this method on this thread before, but I might as well do it now. The method utilises CP during FB just as Briggs 1 and 2 both use and is similar to Roux, but the main thing about this method is that to know it in its entirety you need 28 <RU> algorithms, with one of the algorithms simply being an R and the longest being 11 moves. The steps are as follows:

1. CPFB, 9 moves
2. Triplet at DR. This consists of an E slice edge being placed at DR with 2 corners orientated relative to it, 5 moves.
3. Orientation belt. This is where the algs come in (they require the other E slice edge to be orientated on U), 9 moves.
4. SB while preserving orientation. This uses <R2MU>, 8 moves.
5. LSE, exactly the same as Roux, 13 moves.

Average movecount≈44. Please tell me if any of my numbers are incorrect.
You can find the algs here.
In terms of speedsolving viability, it's a solid maybe. I don't see anyone getting fast with it or bothering to learn it and I also don't see people pushing it to its limit, but I like it.

 

[efattah]

I have generated, tested and optimized the algorithms for twisty ortega to resolve an extremely irritating case for 2x2 & LMCF where you have a T-case on the D-face and want to orient the D face and U face. In color neutral ortega you can usually find an easy face to make on D, but it is hard to be color neutral in LMCF and when forced to create a face from this situation it takes 5-6 moves just to get the face, where it actually takes only 6.71 moves to go from this situation to orienting *both* faces.

Run the inverse of each alg on a solved cube to clearly see which case it solves. I originally created the top 8 algorithms for each case, learned them all, timed them over a hundred reps, and chose the best one. All are easily sub-1.

U-case L2 F R2 B' R2 [5]
T-case R' U2 R2 U' R2 [5]
Sune-case R' F D F' R U' R [7]
Antisune-case R B' U' R' U' R B' [7]
H-case R U' R U' R B' R' U R [9]
Pi-case R' U' R U' R U' R2 [7]
L-case R B U' B2 R' U R [7]

Average 6.71 moves

 

[Hazel]

I have generated, tested and optimized the algorithms for twisty ortega to resolve an extremely irritating case for 2x2 & LMCF where you have a T-case on the D-face and want to orient the D face and U face. In color neutral ortega you can usually find an easy face to make on D, but it is hard to be color neutral in LMCF and when forced to create a face from this situation it takes 5-6 moves just to get the face, where it actually takes only 6.71 moves to go from this situation to orienting *both* faces.

Run the inverse of each alg on a solved cube to clearly see which case it solves. I originally created the top 8 algorithms for each case, learned them all, timed them over a hundred reps, and chose the best one. All are easily sub-1.

U-case L2 F R2 B' R2 [5]
T-case R' U2 R2 U' R2 [5]
Sune-case R' F D F' R U' R [7]
Antisune-case R B' U' R' U' R B' [7]
H-case R U' R U' R B' R' U R [9]
Pi-case R' U' R U' R U' R2 [7]
L-case R B U' B2 R' U R [7]

Average 6.71 moves

These look useful for 2x2, thanks for making them!

 

[RadicalMacaroni]

Not sure if a method like this already exists, but here's an idea for a 3x3 method I came up with:
1. Solve the corners and align the centers (if not already aligned)
2. Put all the edges that belong in the E slice in there, not worrying about orientation or permutation
3. Permute E slice edges
4. Solve top and bottom layer edges
5. Orient E slice edges

 

[Deleted member 36445]

that method is invented 3 times a second

 

[RadicalMacaroni]

I have been experimenting trying to solve a 3x3 one side at a time, like most of us did when we first tried to solve the cube. I did a couple solves and I now average sub 1:30 with a 45.76 single. It's an interesting method. Terrible, but interesting.

 

[shadowslice e]

I have been experimenting trying to solve a 3x3 one side at a time, like most of us did when we first tried to solve the cube. I did a couple solves and I now average sub 1:30 with a 45.76 single. It's an interesting method. Terrible, but interesting.

That's why it's called the fail method

 

[Electrical]

I just had a random thought. What if there was a method for 2x2 where you orient all the pieces in one algorithm (opposite faces solved), then permute all the pieces in one algorithm? This is basically the Ortega method with step 1 and 2 combined. Does a method like this already exist? Or would it just be too complicated?

 

[efattah]

It has been thought of before, but it has two problems:
1. Recognition on step 2 is really really bad and hard
2. There are a huge number of cases for step 2 and the algorithms aren't friendly

Having said that, it might be worth a fresh look, and see if you can create a recognition method; probably the orientation algorithm would have to avoid changing the permutation of the pieces in order to 1-look the solution. But that will increase the length of the algorithm.

 

[Wish Lin]

I support @efattah 's post. This method an be really good if and only if there are good, pure OBL algs so that one look is possible, or else this could no way stack up against other methods because it will be slower and with a higher alg count(Ortega for example)

But if you do have pure OBL and this would probably be the easiest method to learn one looking out there.

 

[Billabob]

As far as I know this idea has not been explored much due to the problems listed above. There is a wiki page on it: https://www.speedsolving.com/wiki/index.php/Guimond_Method

You might find the external links useful.

 

[ezings]

I believe that this is the closest thing to it https://www.speedsolving.com/wiki/index.php/SOAP_Method

 

[Lpell]

A few days ago ,i a site with algs for 2 gen WV,so if you are using 2GR you will remain with EPLL.But i have an idea:if we phase the edges (to have less cases) without inserting the slot,we can use algs to permute the edges and orient corners with LS and get LL skip!

 

[Deleted member 36445]

It has been thought of before, but it has two problems:
1. Recognition on step 2 is really really bad and hard
2. There are a huge number of cases for step 2 and the algorithms aren't friendly

Having said that, it might be worth a fresh look, and see if you can create a recognition method; probably the orientation algorithm would have to avoid changing the permutation of the pieces in order to 1-look the solution. But that will increase the length of the algorithm.

you mean step 1 right? step 2 is just pbl