The New Method / Substep / Concept Idea Thread

[mDiPalma]

Another thing I want to point out is that I said the method is potentially sub-40 moves, not sub-40 in its current state.

idk y all yall r so obsessed w/ makin <40-move methods. it's like ur tryna make a giraffe by hitchin a foal's neck to a instron

petrus w/ optimal CN 3x2x2 & zbll is 35.68 htm (7.99+4.96+9.90+12.08+.75)

yall needta shift loadbearing walls - not plaster drywall, if u know what i mean

 

[Hazel]

idea:
1: Roux FB
2: BRBlock (2x2x1)
3: EOLS
4: DFDB while permuting LL edges (and preserving EO)
5: COLL but with algs that preserve EP

 

[Teoidus]

Approximate movecount: 7 FB + 8 sq + 8 EOLS + 10 L6EP + 13 L4C = 46

 

[Arc]

idea:
1: Roux FB
2: BRBlock (2x2x1)
3: EOLS
4: DFDB while permuting LL edges (and preserving EO)
5: COLL but with algs that preserve EP

I'm not quite sure what your vision for EOLS here is. How are you thinking it would look? Step 4 is literally just LSEP (steps 4b and 4c of Roux). Step 5 is called L4C and it is 84 algs rather than 42 because of "H perm parity" (corners and edges both solved but not aligned)

 

[Abram Lookadoo]

new method-boepll

1)create pseudo block+cp
this is just a 1x2x3 block except the edges on the equator, above the 1x1x3 block, must match a non-up/down center+cp
2.eo
using r,R,U,M orientate all edge pieces
3.ES+Dco
place the FR+BR edges in the equator, while orienting the corners that will go in DRF+DRB
4.pF2L
permutate (solve) the first 2 layers (including equator) using R2,r2,M2,U,u,E
5.2gll
a set of 84 algorithms to finish the solve

 

[Thermex]

new method

1)create pseudo block+cp
this is just a 1x2x3 block except the edges on the equator, above the 1x1x3 block, must match a non-up/down center+cp
2.eo
using r,R,U,M orientate all edge pieces
3.ES+Dco
place the FR+BR edges in the equator, while orienting the corners that will go in DRF+DRB
4.pF2L
permutate (solve) the first 2 layers (including equator) using R2,r2,M2,U,u,E
5.2gll
a set of 84 algorithms to finish the solve

This is a creative idea, but I think the recog is difficult and for the most part it's better to just stick with a regular first block and second block as opposed to psuedo-ones. Right I'm really into EORoux variants; I feel like starting with FB, doing BR square and EOStripe in whatever order is best and finishing with any EOLS method is one of the best approaches to solving. I average around 23 moves/10 seconds with these steps and will probably devote the next week or so to finding a fast, efficient and sub-300 alg 2LLS+LL method.

 

[efattah]

Although I'm less concerned with move counts these days, I would say the only really accurate way to measure move counts is to reconstruct 20 official solves from one particular person and average the move count (even averaging an Ao5 from one competition would give some idea). This guarantees real inspection time, max TPS, etc., and real speed solving conditions. With most methods if you slow down you can cut 5 moves off the solution by seeing things you would have missed at full speed.

I'd be curious to see the reconstructed move count averages from recent competitions from top solvers like Alex Lau, Kian, Max Park, Feliks, etc., and maybe also Anthony Brooks as he uses full ZBLL in competition.

As it stands I highly doubt any person has achieved 44 moves or less average in actual competition solves.

 

[Teoidus]

Alex Lau's solve are available on cubesolv.es.

Last I checked, Roux usually gets around 49 moves, CFOP 58, and ZZ sans ZBLL around 53-58.

 

[gogozerg]

Roux is just CF but you blockbuild the edges together to reduce to MU (why? Minh Thai was doing just fine with CF)

It's just another "edges last" approach, like all the Corners First derivatives.
(I would love to see modern cubists use Corners First in competitions!)

 

[Pyjam]

(I would love to see modern cubists use Corners First in competitions!)

You have likely met one, haven't you? Maybe you meant: from the young generation?

 

[Hazel]

another idea, hopefully this is better than my last few:
1: Petrus block
2: 2x2x1 in RB while solving EO
3: CPLS
4: 2GLL

90 algs total (84 for 2GLL, 6 for CPLS)

 

[gogozerg]

You have likely met one, haven't you? Maybe you meant: from the young generation?

I meant sub-10 competitors. But I may have missed something.

 

[Thermex]

It's just another "edges last" approach, like all the Corners First derivatives.
(I would love to see modern cubists use Corners First in competitions!)

Eric fattah, the person who posted a couple of posts above you, uses his own CF styled method called "LMCF" and averages around 13 seconds with it I think.

another idea, hopefully this is better than my last few:
1: Petrus block
2: 2x2x1 in RB while solving EO
3: CPLS
4: 2GLL

90 algs total (84 for 2GLL, 6 for 2GLS)

This is very similar to the ideas I've been working on (the first three steps, that is), but instead of solving two Roux blocks and then doing EOStripe, you're doing a 2×2×3 which is much more difficult imo. I would just stick with a 1×2×3, BRSquare and EOstripe, it's about the same amount of moves and much easier to perform. The end is pretty good, though I'm not sure how easy pair insert+CP recog would be. Overall I would guess this is probably around 44-45 moves which is great for 90 algs, and about as good as ZBLL.

 

[Sue Doenim]

Hey guys. Sorry I haven't been participating in all of the revamped revolutionary method talking stuff, but here are my opinions on the matter. This thread is the wall we throw things at and see what sticks, so in here, there are no bad ideas. Please, chuck bad things at it. It's fun to throw things like lightbulbs at walls, even though they won't stick. It's fun to watch them explode. This is not, however, the place you will want to fully develop a method. As far as CP first methods go, I think the best you can do is start with a 1x1x3 and go into either a Roux (CPRoux) or ZZ (Briggs/2GR) finish. I think that Briggs, in the form of CP1x1x3-CPFB-EOStripe, fixes the inspection-heavy problems of 2GR. I've never been particularly good with inspection, so this is my belief. Please, feel free to prove me wrong. Otherwise, CP-first is really just not practical. As far as EO first goes, ZZ is kind of the best option, as usually you can't get more efficient than block building, which is the core concept of ZZF2L. LXE (last [unspecified number] edges) methods are promising. With Roux, L6E is tried and true. L7E is kind of a work in progress, and so far, formats for such methods (e.g. 42 and WaterRoux) are FB-solve corners and 2 edges-L7E. in WaterRoux, the middle step is done by solving corners then 2 edges. In 42, you solve a 2x2x1 (2 edges and a corner) and a corner (basically, the oriented U-layer corner is solved, because it can be AUFed to its spot), then solve the remaining four corners. What I think is an interesting concept would be neutrality between these two steps (Originally, WaterRoux was 2x2x1-TCMLL, very similar to 42.). The most needed development is a good L7E method. What would be helpful is an optimal L7E movecount, so we know how to gauge our 17 move agerage with current methods. L8E, as in ECE and SSC, is pretty well developed. EO (in SSC this is done earlier), reduce to L6E, reduce to L4E ("4c state"), solve. Bope style is largely unexplored; I think the best thing might be to solve it LMCF style. I'm not sure where Waterman fits into the LXE scale. An idea I had recently with Waterman is to leave one first-layer edge out, and use the empty slot to help later on, when solving R-layer edges, then solve it. Waterman might deserve more attention than it gets; I'll look into it more. CF methods are technically L12E methods, and I think LMCF is the best way to do this. CFOP (or CFCE [or 3CFCE]) is basically the epitome of LBL. Maybe not. Columns is generally bad. Once again, maybe not. M-CELL is an interesting method, and I think we should look at more ways of solving this. Same goes for the Franciscoes, most notably Hexagonal (which I now noticed are M-CELL [mind blown {I'm using a lot of parentheses and brackets and junk}]). I can't think of other methods at the moment, so okay. Please, prove many of my points wrong. PLZ!!!111!11!!!!!!!!!WEag1DSgafjslidjf!!!!!111!!11!

Here are a couple of method ideas: SSC reduction to 2GLL
This is more of a bad explodey idea.
1) EOEdge
2) Orient D-corners on D-face, solve equator
3) Solve D corners and permute U-layer corners
4) Solve D-layer edges
5) 2GLL

Other style M-CELL/HF building
1) FB+UBL corner
2) Solve M-slice
3) z rotation; solve rest like normal

 

[Hazel]

@Thermex
Well, what about this:
1: Roux block
2: 2x2x1
3: EOStripe
4: CPLS
5: 2GLL

I think the advantage to this over starting with CPFB is that you don't have to rely on inspection as much, you only have to plan out a standard Roux block, then hopefully use the remaining time to start planning out the 2x2x1 instead of worrying about CP. Plus, people who aren't paticularly experienced or good at tracing/planning a lot during inspection will probably have a really difficult time planning out CPFB.
CPLS can take some time to recognize, but recognition can improve by figuring out faster methods of it, and the algs are already really fast. 2GLL is, of course, very fast already.

 

[Arc]

@Thermex
Well, what about this:
1: Roux block
2: 2x2x1
3: EOStripe
4: CPLS
5: 2GLL

I think the advantage to this over starting with CPFB is that you don't have to rely on inspection as much, you only have to plan out a standard Roux block, then hopefully use the remaining time to start planning out the 2x2x1 instead of worrying about CP. Plus, people who aren't paticularly experienced or good at tracing/planning a lot during inspection will probably have a really difficult time planning out CPFB.
CPLS can take some time to recognize, but recognition can improve by figuring out faster methods of it, and the algs are already really fast. 2GLL is, of course, very fast already.

I think doing CPLS before EOStripe is better. There's no benefit to waiting and there is benefit to doing it sooner (algs don't have to preserve EO/DFDB, don't have to worry about FR being oriented)

Also you guys seem be have the wrong idea about CPLS. Traditional CPLS has the edge solved first and is around 35ish algs. Full CPLS, that is, not solving any part of the last pair first, is around 110 algs for ZZ, and more than double that for Roux, though I suppose you could reduce it to the 110 in a maximum of 2 moves.

The whole idea behind classic CPLS is how easy it is to solve that edge in ZZ-style F2L. Usually taking 1 move, with a maximum of 3.

Also as far as recognition goes I outlined what I believe to be a pretty good system in the CPLS thread.

 

[shadowslice e]

Hey guys. Sorry I haven't been participating in all of the revamped revolutionary method talking stuff, but here are my opinions on the matter. This thread is the wall we throw things at and see what sticks, so in here, there are no bad ideas. Please, chuck bad things at it. It's fun to throw things like lightbulbs at walls, even though they won't stick. It's fun to watch them explode. This is not, however, the place you will want to fully develop a method. As far as CP first methods go, I think the best you can do is start with a 1x1x3 and go into either a Roux (CPRoux) or ZZ (Briggs/2GR) finish. I think that Briggs, in the form of CP1x1x3-CPFB-EOStripe, fixes the inspection-heavy problems of 2GR. I've never been particularly good with inspection, so this is my belief. Please, feel free to prove me wrong. Otherwise, CP-first is really just not practical. As far as EO first goes, ZZ is kind of the best option, as usually you can't get more efficient than block building, which is the core concept of ZZF2L. LXE (last [unspecified number] edges) methods are promising. With Roux, L6E is tried and true. L7E is kind of a work in progress, and so far, formats for such methods (e.g. 42 and WaterRoux) are FB-solve corners and 2 edges-L7E. in WaterRoux, the middle step is done by solving corners then 2 edges. In 42, you solve a 2x2x1 (2 edges and a corner) and a corner (basically, the oriented U-layer corner is solved, because it can be AUFed to its spot), then solve the remaining four corners. What I think is an interesting concept would be neutrality between these two steps (Originally, WaterRoux was 2x2x1-TCMLL, very similar to 42.). The most needed development is a good L7E method. What would be helpful is an optimal L7E movecount, so we know how to gauge our 17 move agerage with current methods. L8E, as in ECE and SSC, is pretty well developed. EO (in SSC this is done earlier), reduce to L6E, reduce to L4E ("4c state"), solve. Bope style is largely unexplored; I think the best thing might be to solve it LMCF style. I'm not sure where Waterman fits into the LXE scale. An idea I had recently with Waterman is to leave one first-layer edge out, and use the empty slot to help later on, when solving R-layer edges, then solve it. Waterman might deserve more attention than it gets; I'll look into it more. CF methods are technically L12E methods, and I think LMCF is the best way to do this. CFOP (or CFCE [or 3CFCE]) is basically the epitome of LBL. Maybe not. Columns is generally bad. Once again, maybe not. M-CELL is an interesting method, and I think we should look at more ways of solving this. Same goes for the Franciscoes, most notably Hexagonal (which I now noticed are M-CELL [mind blown {I'm using a lot of parentheses and brackets and junk}]). I can't think of other methods at the moment, so okay. Please, prove many of my points wrong. PLZ!!!111!11!!!!!!!!!WEag1DSgafjslidjf!!!!!111!!11!

Here are a couple of method ideas: SSC reduction to 2GLL
This is more of a bad explodey idea.
1) EOEdge
2) Orient D-corners on D-face, solve equator
3) Solve D corners and permute U-layer corners
4) Solve D-layer edges
5) 2GLL

Other style M-CELL/HF building
1) FB+UBL corner
2) Solve M-slice
3) z rotation; solve rest like normal

I agree that this wall can serve as somewhere where you can just throw a load of ideas at to see what sticks but at the same time there does need to be some level of filtering on a personal level. Sort of a "is this really new or helpful?" If there is a 1 line rebuttle then perhaps it would be better if a method proposer could look for themselves because it can get extremely tedious to go through all of them and quickly find major major flaws.

In addition, it would be good if the designer tries their utmost to make sure the method hasn't been proposed before if trying to claim credit (though I don't mind so much if it's an old concept that they want to discuss in a new way).

Lastly, there does also need to be some filtering as a method designer at some point needs to think "is this really new or just low hanging fruit that someone probably thought of before?" as there are many many things which people will almost certainly have thought of in the past which they never proposed publicly as there was that quite easy to find flaw. To finish, if there is an old method which hasn't been developed mucheck and after loads of thinking you can't work out why, I would be fine with someone bringing it up and seeing what others think (as long as it is phrased in such a way to indicate that).

See my method development primer

 

[Teoidus]

Hello,

I've searched throughout the forums as per @shadowslice e 's suggestions and I couldn't find this method proposed anywhere else, so I've decided to call it TERM1 (Teoidus's Elegant Reduction Method for the 1·1∗1 Rubik's Cube):

Though most methods for solving the 1⊗1×1 Rubik's Cube involve haphazardly flailing the cube about its 3 axes, I have developed an efficient solution for each possible cube state. Because a significant amount of work went into generating (by hand!) solutions for all possible 1x1*1 Rubik's Cube(TM) (R) (C) states, I would like teach a university course in mathematics on this material. It would be greatly appreciated if someone could please point me to someone to contact in order to publish a textbook for this course and offer some advice as to securing a professorship (preferably at Oxford or Cambridge so I can be one of @shadowslice e or @TDM 's superiors).

If I can secure proper means by which to accomplish this I will begin work shortly on TERM2.

Best,
Sincerely,
Warm Regards,
Teoidus, creator of TERM1, a.k.a. Teoidus's Elegant Reduction Method for the 1X1⊕1 Rubik's Cube

 

[efattah]

A brief update on LMCF development with my recent strategy of first finding ergonomic sequences then trying to apply them. This is resulting in far less M-U algorithms and more R/F/U algorithms which have higher TPS.

E2L quadruplet example (sub-1 execution)
[M2] R' F R2 U' R2 F2 R2 U' R2 F
DF->UR, UR(d)->FR, UL(d)->FL, FR(o)->UL

E2L triplet converted to quadruplet with intuitive F/F' injection during the algorithm, slightly less ergonomic
U M2 U2 M2 U L2 [F'] U' M U [F]

E2L pair algorithm, swap UL and FL, both disoriented:
F' R U M2 U' R' F (0.7 seconds or less)

Also a couple of days ago I had a breakthrough, I got a 9.49 full step where I was unable to see the corner solve in the inspection. Previously all my sub-10's and sub-9's only happened if I could 1-look the corners, but my E2L modifications are speeding up the E2L phase enough that it is now possible for me to get sub-10 (occasionally) even without 1-looking the corners, and this still despite my slow TPS of 3-4. My 'average' is still totally random since I only know a third of the algorithms and so the chance of me making it through the solve and knowing the algorithms for each E2L/L6E case is only around 1 in 9 (1/3 * 1/3 * 1/3). However if I know the cases that come up the solves are very fast.

 

[Neuro]

In L7E, why don't we insert LE while doing EO and then 4B/4C? It should give decent movecounts w/o too many algs. And if you want to go the extra mile, implement L6EP.

I think if you use L7E w/ 42/TCMLL neutrality you *may* get a sub-40 avg movecount with 274 algs (unless you go for algorithmic L6EP, then it goes up significantly). I'm worried about L5C recog but if it can be absolutely nailed then this has amazing potential.

EXAMPLES (AVG=17.2)
16
16
19
18
17

EDIT: I think I'm going to gen a ton of algs and compile all CP Roux based methods that have merit and make a thread. So far I have mine (2GRoux, uses Pinkie Pie) the L5C+LE variant, a L7E variant, the base (uses CMLL/LSE), this unnamed method (CPFB, <RrMU> EO Line, F2L, 2GLL), and one similar to LLOB+ZBLL (CPFB, SB, EODFDB, 2GLL) I'll also include some LSLL variants as necessary but I'd like to give credit to the OG creators of these methods so please respond if you've already invented one of these variants. Thanks!