The New Method / Substep / Concept Idea Thread

[Shiv3r]

so guys, I want to bring this up again, but I'm working on making Meyer method a lot faster. I need some volunteers to help me find what solves which case. it's called OP+CMLL, and what it does is solves Orientation parity at the same time as CMLL. so you get algs like this(for adj swap): F sexy F' U' lucasparity. basically I'm coming up with a bunch of algs that have lucasparity or pure parity or double parity in them so we can skip orientation parity while doing CMLL. there will be some cases we'll need to generate algs for, but most of them can probably be done just with experimentation.
and yes, Meep already did this for OLL, it's called POLL.

here's the algs I have come up with so far, if you have any others please tell me.
https://docs.google.com/spreadsheets/d/1cxz3Scm8WlMHIDLC7nzc0a8RQQCwn38v7G_bsfCySiY/edit#gid=0

EDIT: random cool trick: if you use the Sebastian Weyer PLL parity alg, you can replace the r2 with an l2 and it'll do PLL parity on the bottom.(I realized only recently that actually do it mirrored for some reason idk why)

 

[Neuro]

So this was a solve I got using BFTT that had two possible outcomes. One is what I would do in a speedsolve more than likely (alg/execution-wise) and the other uses a lot of cancellations and a shortened alg to reduce movecount. 1st is 47 and the 2nd is 40.

Scramble: F2 D' F2 L2 F2 L2 U' B2 L2 F2 L D' F2 R U2 B L F D' U2 L2

F L' D' B U R U' B2
M' U r U2 M2
L U' R2 U L' U2 R
R' U' R U' R' U2 R
y U2 R' U' R' U' R U R D R' U' R U D' R U R2 U' R' U2

F L' D' B U R U' B2
M' U r U2 M2
L U' R2 U L' U R U' R' U2 R
y U2 R2 U' R' U L' U2 R U' R' U2 L R U R2 U'

I think someone with the right TPS could make this a really viable method for OH and 2H

 

[Thermex]

@Shiv3r I'll try to find some algs this weekend for OP+-CMLL.

 

[Hazel]

Usually, in Roux solves, L6E is done by first orienting the edges, then solving the UL and UR edges, then the last 4.
What if instead, you did it like this:

Solve 1 First Layer edge while correcting centerpieces (~4-5 moves?)
Solve last First Layer edge while orienting LL edges (14 super easy algs)
EPLL (4 algs)

Example:
L6E scramble:
U M' U2 M2 U M U' M' U' M' U2

solution: U2 M' U2 M2 U M' U' M U' M' U (H perm)

Maybe this is less efficient than traditional L6E, but this may be easier to learn or something, for me it's faster anyway.

 

[Teoidus]

https://www.speedsolving.com/wiki/index.php/L5EOP

 

[Sue Doenim]

Here's an idea I had that's similar to Pinkie Pie. There are kind of 2 versions, first is probably better.
V1
1) F2B-1
2) Insert last F2B edge and place UL and UR on D layer: it's kind of like a phasing and/or VHLS type thing, basically if both UL and UR are on U you orient+phase them, if one is misoriented on D misorient the other, if neither of these cases, set one up; the idea is to make them solvable with an AUF and M move
3) Insert last corner while solving CP
4) Pseudo 2-gen OLLCP
5) L6E
V2
1-2) Same
3) Insert last corner+orient edges
4) COLL
5) L6E
Movecount would likely be higher, but not enormously. Other than that, I don't have much to say.
V1 example solve
Scramble: B2 D' F2 D R2 F2 D2 F2 U B2 U B F2 L2 R D' F' L R F
y2
FB- R U F L2 F' D l2 F' r' M F (11/11)
SBsq- R' U M U M' U' M U2 M2 U R2 U2 r U r2 (15/26)
Thing- R U' R2 F R F' (6/32)
PP setup- U' M' (2/34)
CPthing- R U L' U R' U R U R' U2 L (9/43)
OLLCP- U2 r U M U R' U' r U' r' (10/53)
L6E- U' M2 U M2 U2 M U2 M' (8/61)
Movecount was high, but I'm bad. Like, really bad. Still, seeing that movecount, it's probably useless. I'm not even going to try a V2 example.

 

[FastCubeMaster]

When I first switched from CFOP to ZZ, I was doing a lot of EOCross, because I found block building hard. After my second comp, I was able to transition to full ZZ, and I can confidently say that either CFOP or ZZ is much better than a mix of both.

I know this is a bit late, but thanks for that.
Just making sure, EOcross AND/OR CFOP style F2L (in ZZ) is worse than doing a complete method by itself?

 

[FastCubeMaster]

I'm not sure if joke sub-methods count, but it's still an idea.

I want to propose JLL, which isn't an acronym for anything. JLL forces a PLL skip by doing a bunch of PLL algs before OLL, and after the correct state is found, you do the OLL and it's solved.

Haha I know this is a joke method but this is probably one of the dummest things I've ever read.
Thanks for the laugh

 

[TDM]

Here's an idea I had that's similar to Pinkie Pie. There are kind of 2 versions, first is probably better.
V1
1) F2B-1
2) Insert last F2B edge and place UL and UR on D layer: it's kind of like a phasing and/or VHLS type thing, basically if both UL and UR are on U you orient+phase them, if one is misoriented on D misorient the other, if neither of these cases, set one up; the idea is to make them solvable with an AUF and M move
3) Insert last corner while solving CP
4) Pseudo 2-gen OLLCP
5) L6E
V2
1-2) Same
3) Insert last corner+orient edges
4) COLL
5) L6E
Movecount would likely be higher, but not enormously. Other than that, I don't have much to say.
V1 example solve
Scramble: B2 D' F2 D R2 F2 D2 F2 U B2 U B F2 L2 R D' F' L R F
y2
FB- R U F L2 F' D l2 F' r' M F (11/11)
SBsq- R' U M U M' U' M U2 M2 U R2 U2 r U r2 (15/26)
Thing- R U' R2 F R F' (6/32)
PP setup- U' M' (2/34)
CPthing- R U L' U R' U R U R' U2 L (9/43)
OLLCP- U2 r U M U R' U' r U' r' (10/53)
L6E- U' M2 U M2 U2 M U2 M' (8/61)
Movecount was high, but I'm bad. Like, really bad. Still, seeing that movecount, it's probably useless. I'm not even going to try a V2 example.

Yeah this is really inefficient. In general, 2-gen cases are no better than the average case. Also, you're not really saving any moves by doing the edge and corner separately.

 

[Neuro]

I think listing all the things that'd be useful to know for LSE would look like:

Established: Some (or all) of Pinkie Pie i.e. OLLCP, EOLR (the good cases)

Plausible: LMCF style L5E, L5EP/swapped L5EP (and y2 mirrors?), sandwich orientation for when URUL are solved (very rare), and ELL/Opp Cross ELL.

Standard LSE is very well established right now. What I would do is focus on CMLL variants or non-matching blocks
friendly LSE recog.

@JTay hows TOLS coming along?

EDIT: I found a really cool LSLL method for CFOP, looks pretty good for OH. Basically put the LS pieces in with no regard to orientation and 1 alg reduces it to a 2GLL state. Link

 

[Shiv3r]

Hey, so I came up with a 4x4 method a while back, just for the lols, here it is again in more detail. this is based off one of the solves in felik's former WR average, which is basically Pootris on the E slice instead of the M slice.
Here's the steps:
1.2 opp centers
2.Roux block
3.L4C
4.rotate so Rouxblock is on D, and solve one edge, place at DF, then do 3-2-3
5. Make sure centers are a Uw2 away and then keyhole in the RB and LB edges, then solve the DF cross edge
6. Finish as in Yau5(Double Xcross)
it's kind of a stupid method, but hey, it's kind of fun, right?

Here's a 156 OBTM example solve:
scramble: U L2 F U' R Fw' B' L' F' D U' L F U' Uw2 B' Uw2 Rw F L' U2 R2 B Rw' R D R' D' L B' R' D' F' B2 U' Fw D' B L' R
y//inspection
F2 Rw F Rw'//first center
R' y Uw2 U' Rw U2 Rw'//first 2 centers
z' F R L' B D' U' Rw U F2 Rw2 x F2 3Rw' F//Roux block
U 3Rw Rw' U Rw U' 3Rw2 Rw' U' Rw U Rw2 U2 Rw2 3Rw' U' Rw U' Rw' U2 2L' U' 2L//L4C
z' y' F2 Uw' R U R' Uw F'//rotate + any edge at DF
Uw' F' U2 F L' U2 L Uw 3Uw2
R U2 R' Uw F' L F L' Uw'
L' U2 L Uw' F' U F Uw//3-2-3 edgepairing
R U' R' 3Uw2 L' U L F' U' F U R U' R' 3Uw2 L' U' L F2//Keyhole in back F2L edges + last cross edge
U F' U2 F y' U' R' U R
R U R' U' F' U2 F U2 F' U F//finish F2L
U r' U2 r' U2 l U2 r' U2 r U2 x U2 r2 U2 r' U2 r U2
R U2 R2 U' R2 U' R2 U2 R y2//OLL+double parity
R2 3u' R U' R U R' 3u R2 3f R' 3f'//PLL

thoughts on this method?

 

[Hazel]

So I came up with a LSLL method for 3x3, not sure how viable it would be:
1: Insert LS edge while orienting LL edges
2: Solve LS corner while solving LL corner orientation (like an advanced Winter Variation
3: PLL

 

[mDiPalma]

So I came up with a LSLL method for 3x3, not sure how viable it would be:
1: Insert LS edge while orienting LL edges
2: Solve LS corner while solving LL corner orientation (like an advanced Winter Variation
3: PLL

https://www.speedsolving.com/wiki/index.php/MGLS

 

[Hazel]

What about a super alg-heavy version of ZZ-CT for CFOP where TSLE solves all LL orientation, not just the corners?

 

[Shiv3r]

What about a super alg-heavy version of ZZ-CT for CFOP where TSLE solves all LL orientation, not just the corners?

TOLL. already thougt of by Justin taylor and efattah.
this is why I usually don't propose LS methods often, they're usually already a thing.

 

[JTay]

@Neuro I have generated all 116 algs for TOLS+ and TOLS- at this point. I need to make the images for the "-" cases and that's about it. Then, I might tackle improving TTLL.

I considered developing algs for Tran Style 3rd Slot (I believe that's the name), but it's too situational to be used effectively.

If someone feels like learning an absurd number of algs, full Twisted VLS should be exactly 650 algorithms. I believe TS3S with edge orientation is also the same number of algorithms, just a different approach. You could have some pretty insane solves with this though. XRibbon - -> One Pair - -> TS3S/TVLS - -> TTLL/PLL.

Just spitballing some thoughts I've had, let me know what y'all think!

 

[GenTheSnail]

I know this is a bit late, but thanks for that.
Just making sure, EOcross AND/OR CFOP style F2L (in ZZ) is worse than doing a complete method by itself?

yep.
Vanilla ZZ or vanilla CFOP.
Don't mix EO with CFOP unless your're doing some LSLL influence stuff.
Don't do cross-stuff with ZZ unless, like, the pieces solve themselves or something weird.

 

[Teoidus]

I think the efficiency of full blockbuilding is severely overrated, actually

Here are some data from way back when people did movecount analysis on eocross vs blockbuilding: http://www.speedsolving.com/forum/threads/zz-speedcubing-method.5180/page-6#post-69583

If the eocross is completed naively (that is, EOLine and then DLDR), a stepwise optimal pair-by-pair f2l yields ~25.74 moves to complete f2l from eocross.
By comparison, EOLine + stepwise optimal 2square 2pair f2l yields ~24.64 moves to complete f2l from eoline.

That's ~1.1 extra moves more than stepwise optimal blockbuilding + ~3.47 extra to complete the eocross naively.

So in this case there is a ~4.57 move discrepancy, but in practice eocross will never be completed naively and we are much more likely to reach stepwise optimal pair-by-pair than stepwise optimal blockbuilding.

On top of this, eocross f2l is going to be more ergonomic than blockbuilding f2l, as every pair solution has "wrist turning number" 0.

I believe gyroninja has also done analysis which reveals that eocross f2l is actually more qtm optimal than blockbuilding f2l (though this goes away when you allow openslotting): http://cubic-intelligence.tumblr.com/post/149477762423/wow-eocross-gives-lower-qtm

 

[Neuro]

Joking about ZZ Magic !?!?! ITS BEAUTIFUL!!!!!!

We all know ZZ Magic right? the forced LL skip by solving the LR Faces??? Well.... Here's a new way to make it viable!!!

While putting in your LS, solve a 1x1x3 bar on U.... Then... use one of 17 algs to solve the other bar and BOOM. LL DONE!!!

All jokes aside, if you solve a 1x1x3 bar on the U layer, you reduce ZBLL sets to just T and U ZBLL each of which have only 6 algs, and PLL which has 5. I wouldn't ever use this because finding a bar can be hard as well as inefficient. Some people might find it interesting. Oh and it removes the magic aspect b/c you so an LL alg to get an LL skip

 

[Pyjam]

Too difficult. It's already too difficult to build and preserve a single pair on U before inserting the last pair. This would reduces the number of ZBLL cases but phasing is an easier way to reach the same goal.