The New Method / Substep / Concept Idea Thread

[Neuro]

Random idea that was super easy to complete: ZZ-KA (Kristopher Asis.) Probably trash but it's interesting nonetheless

1: EO pseudo line: Line can either be solved OR be solvable w/D2
2: ZZF2L: Everyone knows this
3: COLL: Again, everyone knows this
4: EPLL/OCEPLL: Biggest difference: solves all edges including opposite DB/DF

Total of 51 algs, but probably only need to learn 5 (assuming full COLL/EPLL is known)

Helps in bad EO line cases (and OH if experienced w/ M slice), but probably not any more efficient than fixed COLL/EPLL. I already have a sheet with all of the algs. If nothing else, it's a printable COLL sheet:
https://docs.google.com/spreadsheets/d/1dZUdoMu27MP39NmQwBFUG6Xtg21AvHSkaXpvXvubJ0U/edit?usp=sharing

Credit to Kristopher Asis, he already had OCELL algs made. Here's the website if interested: http://meep.cubing.net/about.html

 

[Sue Doenim]

Random idea that was super easy to complete: ZZ-KA (Kristopher Asis.) Probably trash but it's interesting nonetheless

1: EO pseudo line: Line can either be solved OR be solvable w/D2
2: ZZF2L: Everyone knows this
3: COLL: Again, everyone knows this
4: EPLL/OCEPLL: Biggest difference: solves all edges including opposite DB/DF

Total of 51 algs, but probably only need to learn 5 (assuming full COLL/EPLL is known)

Helps in bad EO line cases (and OH if experienced w/ M slice), but probably not any more efficient than fixed COLL/EPLL. I already have a sheet with all of the algs. If nothing else, it's a printable COLL sheet:
https://docs.google.com/spreadsheets/d/1dZUdoMu27MP39NmQwBFUG6Xtg21AvHSkaXpvXvubJ0U/edit?usp=sharing

Credit to Kristopher Asis, he already had OCELL algs made. Here's the website if interested: http://meep.cubing.net/about.html

I actually had an idea very similar to this today, but with CFOP, where I think it would be more beneficial, due to having 4 cross edges instead of 2 line edges.

 

[AlphaSheep]

I actually had an idea very similar to this today, but with CFOP, where I think it would be more beneficial, due to having 4 cross edges instead of 2 line edges.

It's most beneficial for ZZ with COLL and <M, U> EPLLs because correcting the line edges can be cancelled into the every EPLL so they never take any extra moves.

 

[Shiv3r]

guys so what I have been doing is playing around a LOT on 5x5. I have gone through a lot of methods, from K4(K5?) to OBLBL(one of my favorites still) to Hoya5(better than Yau5 imo), so even Lewis5(where in L5E you solve all the cedges first with L5EO and EPLL algs, then use comms on the wings). My favorite method so far is freeslice. However, when looking at this hoya variant, I think it could have serious potential.

 

[Shiv3r]

It's most beneficial for ZZ with COLL and <M, U> EPLLs because correcting the line edges can be cancelled into the every EPLL so they never take any extra moves.

but I think that the extra move makes it useful, as you can also cance before doing LL by doing a D2 before ZZF2L. 5 or so algs to basically do a D2 which isn't too hard
but let's take this method Idea even further.
it may be much much better if the two edges to be swapped are at FR and RB, because then you can make much more efficient pairs. And add another 4-5 algs, and you can do psuedopairs on both sides as well. come up with an acronym for this algset, and then there will be 2 sets: xyz-1, which solves EPLL with FR and BR swapped, and then xyz-2, which does it with both sides psuedo-pairs. Im calling it LELL for now, for Lewis fix E-slice edges & Edges(2 parts for one "E") of LL.

I happen to love psuedopairs, and I get them all the time in Roux solves, and theyre even easier to make in ZZ. the problem is fixing them. Alg count would be 4(EPLL)+ 5(LELL-1)+5(LELL-2, one of which is z U2 M2 U2 M2) = 14 algs, you should already know 5 of them so only 9 algs to learn for a faster and stronger ZZF2L, which is the weakest step in ZZ according to most ZZ solvers.

can Neuro help me(since he brought up ZZ-KA)?

 

[Neuro]

I'll take a look at it and see. All the algs would be 2 gen and recog is the came as EPLL so it may be nice. But it'll take people a while to grasp being able to swap those 2 while solving. Just speculation though, it's probably not that big of a problem.

 

[efattah]

Just to update, Neuro and I have been working the Roux-Waterman hybrid and it is looking amazingly promising and I don't say that lightly. Right now the sequence looks like this:
1. FB
2. 2x2 Block on back right (partial second block)
3. Insert the DFR corner in random orientation (0-4 moves ultra fast)
4. Solve the corners now with either CLL, TCLL+, TCLL- based on the DFR corner orientation
5. Based on the status of the last 7 edges, you finish the solve with a variable algorithm set based on your preference and how low move count you want
[option 1; solve FR intuitively then do Roux LSE]
[option 2; use Waterman L6E; all cases are listed in LMCF document PDF, revision 4.1 or greater]
[option 3; use expanded Waterman L7E, currently being generated]

Here is an example we made up:
https://alg.cubing.net/?alg=x2_D-_R...D2_U2_R-_D2_L2_R2_U-_F_L-_D_U2_F-_D2_L2_R_D2_

Scramble: B2 R F2 R2 D2 U2 R' D2 L2 R2 U' F L' D U2 F' D2 L2 R D2

x2 D' R2 B L' U x // first block
L2 D' L2 U' F' U' F r' // partial second block
R U R' U F' U2 F // TCLL-
U M U' M2 // solve UL edge
R U r' U M' U' r U' // Waterman L6E Set 1 case 7B
r' U2 M' U2 M2 // permute midges

Total 37 STM full step

This was a pretty favorable solve in terms of move count, I would expect the average to be higher but still lower than Roux.

It also works as a hybrid method where you can finish the second block Roux-style if the case is favorable, otherwise branch into the Waterman variant and insert the DFR corner in random orientation if the second block is unfavorable.

 

[Shiv3r]

can you help me gen LELL algs neuro? I am busy and won't be able to be around a solver for a few weeks(travelling)

 

[Neuro]

Once I'm done genning I'll PM you and if it's good then we can start optimizing.

 

[Shiv3r]

Once I'm done genning I'll PM you and if it's good then we can start optimizing.

Hey, another Lewis method video.

 

[AlphaSheep]

ZZF2L, which is the weakest step in ZZ according to most ZZ solvers.

According to which ZZ solvers exactly? Because to me ZZF2L is the strongest step in ZZ, and I think many ZZ solvers would agree with me.

 

[Y2k1]

According to which ZZ solvers exactly? Because to me ZZF2L is the strongest step in ZZ, and I think many ZZ solvers would agree with me.

I second this, zzf2l is the main reason I use zz over cfop, misoriented edges suck to insert and pair in cfop

 

[Shiv3r]

According to which ZZ solvers exactly? Because to me ZZF2L is the strongest step in ZZ, and I think many ZZ solvers would agree with me.

I second this, zzf2l is the main reason I use zz over cfop, misoriented edges suck to insert and pair in cfop

I said that because someone a few pages back said that the weakest part of ZZ is the ZZF2L. and then they said that the best way to make ZZ better would be to optimize the ZZF2L, the LL and the EOline are pretty good.

Also, if you use LELL then you get a much more efficient ZZF2L because the chance that you will get an unintentional pair is I believe Doubled if I'm not wrong.

Basically if you think ZZF2L is good, this makes it even better, and LELL is only a few algs. you only REALLY need to Learn LELL-1 if you're too lazy to learn all the algs, because for LELL-2 you can execute the EPLL, then do a z/z' and then M2 U2 M2.

I only have the face-turn algs right now, haven't looked at algs that involve the M-slice yet.

 

[efattah]

Okay Neuro and I have officially named the Roux-Waterman hybrid as 'WaterRoux.' He is generating the TCMLL+/- algorithm set and I am generating the L7E set. We should have the algorithm sets done in a week or so. This could really be big.

I would add a wiki page if I knew the procedure...

 

[Miro]

I like this! I would suggest a further refinement:
1. FB on left
2. Solve a square on DR
3. Solve last 5 corners in one step (probably involves an AUF set up move) ignoring FR edge
4. Solve L7E (ordinary LSE slots + FR)

I am generating the L7E algorithms as we speak for the advanced section of the LMCF document and so far the L7E algorithms look amazing. Super ergonomic [R, r, U, M] and the movecount average is looking to be [0-3 setup]+[10.7 move algorithm to solve L/R and orient midges] + [4.5 to permute midges] = 16.7 moves for L7E.

Another idea for this method:
2. Solve square on DR + corner, with swapped corners (RFD, RBD)
3. Solve last 6 corners with EG1-style
4. Solve L7E (ordinary LSE slots + FR)

This way you can implement EG2-style too.

EG0, 1, 2 have better recognition than TCLL. But stranger second, or both blocks. All four D-corners can by freely permuted.

 

[AlphaSheep]

I said that because someone a few pages back said that the weakest part of ZZ is the ZZF2L. and then they said that the best way to make ZZ better would be to optimize the ZZF2L, the LL and the EOline are pretty good.

Also, if you use LELL then you get a much more efficient ZZF2L because the chance that you will get an unintentional pair is I believe Doubled if I'm not wrong.

Basically if you think ZZF2L is good, this makes it even better, and LELL is only a few algs. you only REALLY need to Learn LELL-1 if you're too lazy to learn all the algs, because for LELL-2 you can execute the EPLL, then do a z/z' and then M2 U2 M2.

I only have the face-turn algs right now, haven't looked at algs that involve the M-slice yet.

No, it doesn't make it better, because you're saving a handful of moves in F2L at the cost of adding them back later with worse recognition. The best approach to ZZF2L is free form block building. You need to be able to build the blocks as efficiently as possible, and sometimes that does involve having swapped edges for easy pseudo pairs. I do it when it makes sense, especially during the first block, when I can correct them easily with u2 R2 u2, but theres only a meaningful advantage to that when both pairs are already formed - and that happens maybe in 1 in 50 solves. Not to mention that if you're block building right, many of your pairs won't even involve E slice edges which already massively improves efficiency.

 

[Shiv3r]

No, it doesn't make it better, because you're saving a handful of moves in F2L at the cost of adding them back later with worse recognition. The best approach to ZZF2L is free form block building. You need to be able to build the blocks as efficiently as possible, and sometimes that does involve having swapped edges for easy pseudo pairs. I do it when it makes sense, especially during the first block, when I can correct them easily with u2 R2 u2, but theres only a meaningful advantage to that when both pairs are already formed - and that happens maybe in 1 in 50 solves. Not to mention that if you're block building right, many of your pairs won't even involve E slice edges which already massively improves efficiency.

In fact, I agree with you. I wanted to claim LELL as my own idea(and give it that stunningly amazing title lelelel). also this was mostly based on the Idea that I do psuedoblocks on roux sometimes, and I assumed it'd be easy to do with ZZ. for roux I just do D2 L2 D2 after FB, or for SB cancel into R2 U2 R2 U2 R2.

 

[Shiv3r]

Hey, just came up with a method for 5x5+, maybe even 4x4 if you get really nice centers in inspection
I'm calling it gay Redux cuz it isnt straight redux (at least unofficially-- I'll probably give it a more politically correct name later)

so basically it lets you have the benefits of a Yaucross(thats the technical term, even if it's in Hoya, kind of like how FB is always roux FB no matter what method its in), but still the center freedom of straight redux. its probably a **** method, but maybe it can be useful.
Here's the basic Idea when I came up with it:
-Centers
-pair edges and solve cross dedges, use freeslice
-F2L and then LL

There are two ways to go about with the cross dedges strategy. this is currently what I do:
-Pair up all 4 cross dedges and place them on the D face, try to permute as many of them as possible
-then fix the cross afterwards with R2 U/D R2 moves
-then finish freeslice+L4E like you would in Yau5 or something.


I think it is definitely worse than Straight redux, but I mean it works. I'm not sure if its better than Yau5 or Hoya5 on bugger cubes tho.(more like 6x6-7x7, Yau/Hoya can be decent on 5x5)

 

[efattah]

I finished created the last 7 edges algorithm set for WaterRoux. You can find it in the WaterRoux thread:

https://www.speedsolving.com/forum/threads/the-waterroux-3x3-method-thread.64140/#post-1223644

 

[xyzzy]

I'm not sure if its better than Yau5 or Hoya5 on bugger cubes tho.(more like 6x6-7x7, Yau/Hoya can be decent on 5x5)

Mattia Furlan seems to prefer forming white edges first on very big cubes, albeit without actually forming a cross. (video)

Humans are slow at decision-making, and the difference in efficiency between choosing any edge versus choosing only among white edges is reduced for larger cubes, so this might be a reasonable tradeoff.