The New Method / Substep / Concept Idea Thread

[Sue Doenim]

I've got a pretty crazy idea here. It's a hybrid between LMCF and BLD solving.
0 [inspection]) BLD trace the corner solution, look at edges and plan out first E2L pairs
1) Execute corners using 3-style commutators
2) Execute planned E2L, finish E2L
3) L6E
Insane BLD people can memo in >10 seconds, so 3-4 seconds for corners. Add in 2-3 seconds for picking up cube and starting timer, you have 8-10 seconds for planning E2L things. You should be able to get insane stuff done. 1 looking anywhere from L10-L8E, and even more possibly. Since not a lot moves around during E2L pairs, it should be possible to get a bunch planned out. I'd like to know what you think, especially @efattah .

 

[Hazel]

Proposal for a Roux-variant method called C-Turtle (pronounced like "sea turtle"):
Basically, it's Roux with about 2.5 times as many algs (but I don't think there's even one bad alg), but it solves CMLL while finishing the blocks.
1) C-Turtle
1a- First block, but the DFL corner can be any corner as long as it's oriented
1b- Second block, but the DFR corner can be any corner as long as it's oriented
2) L6C
2a- Insert one D-layer corner with CLS, 102 algs
2b- Insert the last D-layer corner while solving CP, 6 algs
2) LSE (just like in regular Roux)

I call it C-Turtle because a turtle is (no offense to turtles) basically a cute but strange lump with a shell, and after step one, the solved D-layer pieces is basically an odd lump, and the solved E-layer resembles a shell. The C part is because the solved pieces after step one resemble a C shape when the center is also solved, and also because it's a pun (like sea turtle).

 

[Abram Lookadoo]

Proposal for a Roux-variant method called C-Turtle (pronounced like "sea turtle"):
Basically, it's Roux with about 2.5 times as many algs (but I don't think there's even one bad alg), but it solves CMLL while finishing the blocks.
1) C-Turtle
1a- First block, but the DFL corner can be any corner as long as it's oriented
1b- Second block, but the DFR corner can be any corner as long as it's oriented
2) L6C
2a- Insert one D-layer corner with CLS, 102 algs
2b- Insert the last D-layer corner while solving CP, 6 algs
2) LSE (just like in regular Roux)

I call it C-Turtle because a turtle is (no offense to turtles) basically a cute but strange lump with a shell, and after step one, the solved D-layer pieces is basically an add lump, and the solved E-layer resembles a shell. The C part is because the solved pieces after step one resemble a C shape when the center is also solved, and also because it's a pun (like sea turtle).

what if you place the DBL corner in DFR?
edit: DBR in DFL

 

[Hazel]

what if you place the DBL corner in DFR?
edit: DBR in DFL

I meant that you could put any corners in DFR and DFL besides DBR and DBL, because DBR and DBL are needed to make the blocks.

 

[efattah]

I've got a pretty crazy idea here. It's a hybrid between LMCF and BLD solving.
0 [inspection]) BLD trace the corner solution, look at edges and plan out first E2L pairs
1) Execute corners using 3-style commutators
2) Execute planned E2L, finish E2L
3) L6E
Insane BLD people can memo in >10 seconds, so 3-4 seconds for corners. Add in 2-3 seconds for picking up cube and starting timer, you have 8-10 seconds for planning E2L things. You should be able to get insane stuff done. 1 looking anywhere from L10-L8E, and even more possibly. Since not a lot moves around during E2L pairs, it should be possible to get a bunch planned out. I'd like to know what you think, especially @efattah .

So you are suggesting solving the corners in such a fashion that the edges are not affected...? Which allows you to use the inspection time to plan the E2L phase? If so, someone with practice can see two E2L pairs into the future (4 pieces), or if there is a triplet, then 5 pieces, most of the time at least. In fact, interestingly, E2L lookahead difficulties usually make E2L solves significantly sub-optimal. Re-doing a solve much slower I can usually do a way better job of reducing moves & optimizing. So being able to plan E2L during inspection, the savings in terms of move counts would be significant. A few times ever I had the luck of such an easy CLL skip that I could already look ahead into the E2L phase during the inspection, and those instances were my PB singles. I will also say that if I start with a cube that has the corners already solved, I can finish the cube in an astonishingly short time if I am allowed to use the 15 seconds to plan E2L.

 

[Thom S.]

I've got a pretty crazy idea here. It's a hybrid between LMCF and BLD solving.
0 [inspection]) BLD trace the corner solution, look at edges and plan out first E2L pairs
1) Execute corners using 3-style commutators
2) Execute planned E2L, finish E2L
3) L6E
Insane BLD people can memo in >10 seconds, so 3-4 seconds for corners. Add in 2-3 seconds for picking up cube and starting timer, you have 8-10 seconds for planning E2L things. You should be able to get insane stuff done. 1 looking anywhere from L10-L8E, and even more possibly. Since not a lot moves around during E2L pairs, it should be possible to get a bunch planned out.

This is very interesting and I can see much potential. Only question is how much time is lost due to cummutators instead of something like EG

Didn't think much about this one, but how about using your commutators to only solve the 2x2 layer(may end in faster cases, as full 3-Style Corners gives you more options) then E2L and at last a CxLL Set, which solves Part of L6E?

 

[Teoidus]

But,

4 commutators for corners = ~9 * 4 = 36 moves

Compared to ~13-15 moves for EG... I'm sure you can turn faster from planning more in inspection, but can you turn so fast that it overcomes a 20 move disadvantage?

In fact the movecount of this method would be comparable to that of CFOP, and I don't think you'll be able to turn faster than a CFOPper.

 

[Sue Doenim]

One interesting thing is how it would compare to normal BLD. It would definitely be different. For one, your orientation is not locked in by the centers.
You could pick an angle where at least one corner is solved. You could also solve the U layer corners an AUF away from solved. You could, like @Thom S. said, solve D corners and end with a CLL that only affects a few pieces. I'll do an example solve.
Scramble: F R L2 B R L2 F2 L U F U2 B2 L' U2 R2 L' U2 R B2 R
Inspection:
y'// I now solved one of the D layer corners
Memo: TX// I solve 2 D corners
L(Q)// All D corners are solved, shoot to fast U corner position.
CLL will be T 2-gen
R D' L' U2 L D2 L' U2 L D' R'//TX
R' U L' U' R U L U'//LQ
U R U2 R U' R U' R2 U2 R U R' U R//CLL
33 moves, but only 30 with cancellations. You'd probably see them pretty easily. I am kinda losing faith in this though.

 

[Thom S.]

But,

4 commutators for corners = ~9 * 4 = 36 moves

Compared to ~13-15 moves for EG... I'm sure you can turn faster from planning more in inspection, but can you turn so fast that it overcomes a 20 move disadvantage?

In fact the movecount of this method would be comparable to that of CFOP, and I don't think you'll be able to turn faster than a CFOPper.

Valid point.
If we consider that every 3-Cycle can be executed Sub-1, at four commutators, we actually loose only two seconds from a 2x2 expert. The only Question is if higher efficiency and TPS in E2L or higher efficiency and TPS in Corners is worth it

 

[efattah]

Valid point.
If we consider that every 3-Cycle can be executed Sub-1, at four commutators, we actually loose only two seconds from a 2x2 expert. The only Question is if higher efficiency and TPS in E2L or higher efficiency and TPS in Corners is worth it

There is even more to than that. If you don't constrain any corners and leave all of them free, you can force (for example) the blue-yellow edge to always been in the DBL position and then solve the corners in such a fashion that that fixed edge becomes solved, by solving the corners 'around' the edge. I did some random scrambles and found that almost half the time there seems to be two opposing edge pieces that are oriented and aligned. In that case you force them to the D face, solve the corners and you finish the corners with 2 edges solved (50% of the time), or 1 edge solved (every time). With a decent E2L case you can see a pair and a triplet (5 edges) into the future. Which means you can see the entire solve except for L6E, most of the time. I would even venture that in 1 in 50 solves you would be able to see the entire solve in the inspection.

If you consider that even with CFOP, lookahead challenges still limit the solve times (especially averages), so a new method with the same movecount that allows you to see further ahead into the solve would be huge. The corner commutator method would be similar to a CFOP solve where you can see the entire solve except PLL, in the inspection.

Furthermore, if my guess is true that you could see the entire solve in the inspection 1 in 50 times, you could use this method for 3BLD and try for world record singles. You would need to go to around 4 competitions to have a decent chance, but when it happens you would get a 55-60 move BLD solve, which could be finger tricked in 5-6 seconds or even faster.

 

[Spencer131]

Valid point.
If we consider that every 3-Cycle can be executed Sub-1, at four commutators, we actually loose only two seconds from a 2x2 expert. The only Question is if higher efficiency and TPS in E2L or higher efficiency and TPS in Corners is worth it

I think the corners could be done in 3-4 seconds by top bld solvers, but there's still no way to justify spending more than an extra second on corners.

Here is a mediocre idea I had and I dont think its on the lmcf doc. You should be able to at least predict the edges at bl, bd, and dl because 2x2 algorithms can always be done with RUF so that back section of the cube don't be affected.

 

[efattah]

I think the corners could be done in 3-4 seconds by top bld solvers, but there's still no way to justify spending more than an extra second on corners.

Here is a mediocre idea I had and I dont think its on the lmcf doc. You should be able to at least predict the edges at bl, bd, and dl because 2x2 algorithms can always be done with RUF so that back section of the cube don't be affected.

I sort of do this already... when I get an EG-1 case I know the edge that sits between the 'bar' at the BD position never changes, so I can plan to use it in the first E2L pair/triplet. But good point about RUF.

 

[PapaSmurf]

This is an idea for a zz 4x4 with inspiration from ns4 and zz-hoya, although I haven’t tried it out and it could be really bad, but here goes.

1. Solve your F and B colours and place on L and R (so I have blue and green for zz on 3x3, so I’ll solve them into L and R).

2. Solve your line edges and 2 other oriented but random edges into FL and BL.

3. Solve D and B centre and place the FL edge between them. (Also, if possible, pair up another edge while doing so.)

4. Solve two edges. Place one oriented in BR and one oriented in DF.

5. Pair the last 6 edges and do Petrus style eo. OLL parity here.

6. ZZF2L.

7. COLL and EPLL+parity.

Example solve coming tomorrow, when I have time.

Honestly, for all I know this could be groundbreaking or completely stupid and should be buried. Please say if it is either. Also, first method idea!

 

[The Pocket Cuber]

How about this method for solving a 2x2, it may be dumb, but it may be good.....
1. Orient corners average 5 moves
1a) 2 adjacent corners oriented
1b) One out of 72 algorithims
2. Solve corners don't know how many moves
Don't know how many algs, or if it's doable.

There's already a method like this called Sortega except you sort the corners and then do a pbl, which I would enjoy however it is very hard to predict the PBL, while this method you do a bunch of it is easy to predict your last algorithim in inspection. Tell me what you think

 

[Sue Doenim]

How about this method for solving a 2x2, it may be dumb, but it may be good.....
1. Orient corners average 5 moves
1a) 2 adjacent corners oriented
1b) One out of 72 algorithims
2. Solve corners don't know how many moves
Don't know how many algs, or if it's doable.

There's already a method like this called Sortega except you sort the corners and then do a pbl, which I would enjoy however it is very hard to predict the PBL, while this method you do a bunch of it is easy to predict your last algorithim in inspection. Tell me what you think

This method is one of the methods that gets reproposed every once in a while. It's not really viable due to the second step's algcount. More than a few hundred. The HD method fixes these problems. In HD, you orient all corners, like in Guimond, but you force Vs on both sides. Then you use one of 36 algs to solve. Check the link in my signature for some more details.

 

[The Pocket Cuber]

This method is one of the methods that gets reproposed every once in a while. It's not really viable due to the second step's algcount. More than a few hundred. The HD method fixes these problems. In HD, you orient all corners, like in Guimond, but you force Vs on both sides. Then you use one of 36 algs to solve. Check the link in my signature for some more details.

ThankYou!

 

[PapaSmurf]

Example solve coming tomorrow, when I have time.

It is tomorrow, and I have time:
F2 R UW R2 BW L UW’ RW’ F2 RW// F and B centres, 10/10
U R UW2 R’ L’ UW2 U R’ z R’ F2 LW’ RW’ U D’ // Line edges and 2 random and oriented edges, 14/24
R B’ L U’ L F’ RW L’ U’ l2 U2 l2 L D’ L’ // D and B centres and position oriented edge, 15/39
R’ UW’ R U R’ F R’ F’ R UW R2 U2 R2 U2 R2// Solve 2 edges, place at BR and FL oriented. FL to allow for centre building, 15/54
RW U’ RW’ L F’ RW’ F2 RW L’// Solve last two centres and place edge in DF oriented, 10/64
F’ UW’ F’ U2 F UW L’ U’ L R U2 R’ UW’ R U R’ F R’ F’ R UW// Pair edges, 21/85
R U’ R2 F R B’ U2 RW U2 RW’ U2 RW’ D2 RW D2 RW’ B2 RW B2 RW’ D2// Orientate edges + Oll parity. Accidentally placed mis-oreinted edge in BL at the start, 21/106
y’ U’ L’ U’ L R U’ R’ U’ R’ U’ R U’ L2 U’ L U2 L’ U L R’ U’ R U’ R’ U R// ZZF2L, 26/132
U’ R U2 R D R’ U2 R D’ R2 U2 R U R’ U’ RW2 F2 U2 r2 U2 F2 RW2 U R U’ R’// Last Layer, 27/159

Just for reference, this is like my second solve with this method, and I got a 142 mover with yau on the same scramble, so a bit more practice, and it could be possible. Or I'm really inneficient on 4x4. And I also lost the scramble. Sorry.

 

[Abram Lookadoo]

new method (BFLE)

1) create a 2x2x3 block in DB
this step should be seen in inspection (9-15 moves)

2) solve the 1x2x3 block in FD
solve the block using intuition (13-21 moves)

3) LL-eo
use a speed optimal algorithm that solves everything in the U layer ignoring eo (1/496 algs)(7-15? moves)

4) eo
solve eo using a speed optimal algorithm (1/8 skip chance)(1/3 algs)(0-14 moves)(11.25 moves avg)

 

[Sue Doenim]

new method (BFLE)

1) create a 2x2x3 block in DB
this step should be seen in inspection (9-15 moves)

2) solve the 1x2x3 block in FD
solve the block using intuition (13-21 moves)

3) LL-eo
use a speed optimal algorithm that solves everything in the U layer ignoring eo (1/496 algs)(7-15? moves)

4) eo
solve eo using a speed optimal algorithm (1/8 skip chance)(1/3 algs)(0-14 moves)(11.25 moves avg)

Not the best idea. First of all, your style of F2L is not super feasible. One looking a 2x2x3 is not easy. Otherwise we'd have people making XXCrosses every solve. Second, you're using a giant algset to force a tiny, bad set of cases. You're better off with almost any other LL method.

 

[xyzzy]

It's not really viable due to the second step's algcount. More than a few hundred.

Actually, it's "only" about 150 if I'm counting it right (reduced by AUFs both before and after the alg). It's possibly viable, but there isn't much reason to use it over EG.