The New Method / Substep / Concept Idea Thread

[V Achyuthan]

I put a lot of thought into ideas before I decide to post them somewhere. I've thrown away thousands more ideas than I've actually developed. I only develop things that I know are completely unique for the time and or appear to be good for speedsolving. The important thing is to examine every detail of an idea that you have and to not be attached to it just because it is your idea and because you really want it to be good. If it isn't good, then you have to move on to another idea. Keep trying ideas with no self bias and you'll come across one that is objectively good.

If you put an idea out there, expect for it to receive judgement. My judgement is that I don't like that there are pieces in the way for the algorithm-based F2L steps. The same as in Mehta there is an edge in the way of the other pieces. So when you solve the corners the algorithms have to work around or take out the edge. This is as opposed to APB where the first step keeps the pair on the U layer out of the way and the edges are oriented while inserting this pair. Then the L3P step is almost completely free with the RUS moveset available without having to disturb the pair too much. I'm a fan of freedom in methods. But maybe others like the restrictions. Also, try to do something along with the EO step. This will reduce the number of steps and likely also the overall movecount.

thank you

 

[DuckubingCuber347]

TBH. This might be at the same level as APB.

Same level as APB with more steps, less devolvement, no results, etc. I don't think so.

One problem with the APB I face is that you need to learn 116 Algs just to solve 3 pieces (ya I know this can be done intuitively). in this method, the number of algs you need to learn for Corners of D layer is actually really low, since most are F2L algs and also last 2 edges which are DF and FR2 are like not more than 25-30 algs maybe. I have not calculated yet.

Woah there, I hope you aren't implying more algs is a bad thing. There are so many things wrong with that statement. L3P is an intuitive step that is executed algorithmically. I wouldn't think of it as 116 algorithms but instead as 116 cases there are so many trivial cases in that number that can not be justified as an alg. It's so intuitive that the alg spreadsheet is more of a guide leading you through efficiency then telling you "Do this certain set of moves or you'll RUIN THE CUBE DUDE!!". Also I'm pretty sure solving the two corners is eighty algorithms much less intuitive (In terms of solving efficiently) so I have no idea how you got twenty-five. On top of that recognition is roughly equal to L3P because corner recognition is harder than edges.

Also in APB during the EO pair step you have 2 unsolved belt edges, which makes it difficult to recognize EO (although this can be done faster in higher levels). while this method has only one Belt edge unsolved and also FR solved which is actually easier recognition and less algs. and again I don't know if this is better or worse or equal as APB.

Umm... not really. It can be done faster at any level due to several things. I don't know how much you know about APB but you don't look at all the pieces in any given moment during F2L. You first build your Pair (2-4 moves if it's not already done) and once that is done you'll instantly know the orientation based on if the pair can be inserted or not as is. After that you only need to look at five pieces, not six as you're implying since due to process of elimination you can determine whether the one you are not watching (Let's say BR) is a "good" edge or a "bad" edge. Easier recognition and less algs isn't better if the steps before that are worse. That's like arguing that OCLL>PLL is better than ZBLL. Sure, PLL has good recognition but the combined time of recognition for OCLL>PLL is around the same as ZBLL and executing one alg is faster than two (That's why EG is so prevalent in 2x2).

probably could say the same to APB can't you?

That's what I was comparing it to. You can compare it to Petrus, Mehta, and APB since they all start with a 223

Also during higher levels a solver should be able to plan 222 + 223 + FR1 in inspection Ig. Then the rest of the solve would be just algs.

If you can plan 223 + FR in inspection the why would you be lazy be learning less algs for worse movecount?

Mehta is like a damn god method. you can't just go comparing every new method that someone invents with that.

The whole point of comparing new methods to already established methods is to determine if it's viable. If you don't think your method is comparable in terms of potential to Mehta than it's a bad method. Don't try to weasel your way out of things you can't defend because that just shows everyone how little work you put into it and how unprepared you are to show it's viability. If you want to make a method that gets positive attention be able to give results and admit to its weaknesses and try to fix those weaknesses.

ok then. Mehta TDR has 5 steps which are, FB - 3QB - EOLE - TDR - ZBLL. So does APB, 223 - pair - EO while inserting the pair - L3P - ZBLL. And so does this method, 223 + FR1 - Corners of D layer - rotate and do EO - DF and FR2 - ZBLL. and I DIDN'T SAY this method is and IMPROVEMENT.
I JUST SAID I HAVE INVENTED A METHOD. WHY DON'T YOU UNDERSTAND IT DUDE?

Are you claiming that FB > 3QB is two step but 222>223 is one step, although you also claim that FR1 can be planned in inspection also so really 222>223>FR1 is one step which is literally the same thing as FB>3QB.

If it is not an improvement than why are you trying to prove it's viability as a contender for being better or equal to Mehta and APB?

ok. please don't start an argument now. the rest of my method is different right? also it is easier recognition I think. There is just a thing I want to just confirm. When athefre came out with the APB method, everyone said it is good/great/improved version of petrus. but when I come out with any method people just give mean comments. Is it because athefre has given a good explanation and has more experience and I didn't give a good explanation and I have only one year experience? again don't want to start an argument and no offense to athefre or anyone. and definitely not saying APB is bad, I have actually tried it and it is definitely a lot better than petrus and maybe even ZZ.

We were actually not trying to start an argument. I just wanted to know what advantages it had over APB and you gave me some weak points that proved absolutely nothing. When PiKeeper pointed something out you became offended and started being aggressive while spamming posts that had a whiny tone and added nothing to constructive criticism. It helped no one. Athefre's methods get a lot of positive feedback because he pours a lot of work into his ideas before publishing them making sure there are no major flaws and being prepared to answer questions in a responsible, helpful way.

Your methods don't get that same attention because what we get from your methods is that something popped into your head and you constructed it without much thought and then post shortly after an unoriginal, amateurishly put together mess of a "method" that is usually just a worse way of doing another method. You then become offended when people don't believe in it as much as you do and make trashy posts that just create toxicity.

I am not gonna argue anymore. I don't care (no offense) if anyone likes this method or not. I am proud of my invention and that is all that matters to me.

If you are proud of it that's fine but you should know that in real lie this isn't how it works and sometimes you just have to recognize your mistakes and failures. You can't always hide in your own world thinking what you did is justifiable and that everyone is just out to get you because they have some kind of grudge against you. I for one do not have a grudge and try to make things right with people I can be negative to (including Lukascubes).

If you work on refining this method and give proof maybe people would pay actual attention but with how you present your methods most people will continue to take it with a grain of salt.

(Gosh this is a long post. [ @BenChristman1 ])

 

[CornerTwisted]

B: Belt-1
E: EFL
C: CFL and last slot
O: OLL
P: PLL
This is a kind of transitional method from beginners method to CFOP
Belt-1
The belt is an intuitive step, the goal is to get 3 pieces from the middle layer to their designated spots. This step can take around 5 moves if done intuitively and is very easy to plan the whole step in inspection.
Edges of the first layer
This step requires M and U moves to insert each edge on the first layer. To do this, it is essentially making a cross while keeping the belt-1 solved. This takes around 15 HTM moves for me, and you can play around with this and get probably even less moves.
Corners of the first layer+LS
In this step, you use the open slot you have in your belt to insert the corners of the first layer. Then for the last slot you pair the last edge as an F2L slot and insert it. This is really fast and takes somewhere around 20 moves.
OLL and PLL
If you are familiar with CFOP or LBL, you already know this step very well. tThe last layer can be done from 2 look, 3 look, 4 look and anything with more steps, though assuming you know full OLL and PLL, it takes average 25-30 moves, but is super quick as there is barely any intuition, and it is all algorithms.

I tried this method for a bit, and got some sub 20 solves with a PB of 16.83(I average 18 with CFOP and CFCE methods)

I recommend to beginners who are learning F2L but aren’t very fast at it or struggle with their first 2 layers

This is just an alternate F2L strategy
Advanced:
CFOP
BCOP

 

[V Achyuthan]

Just made this doc of 1 scramble 61 methods

1 scramble 61 methods

Scramble B D' F2 D2 L2 D2 R2 B2 L2 F' U2 R2 B' U R' B2 R D' R F L2 CFOP z2 y' R' F R D' R' F2 D' // Cross L U L' U L U' L' // First pair d' U' R L U' L' R' // second pair D U R U' R' D' // third pair R U' R' U' R U R' // fourth pair R' U' F' U F R // OLL U' R U R' U' R' F R2 U' R' U' R U R' F' /...

docs.google.com

Feel free to tell me if I missed any method(s).
Thank you
Edit : These solutions do not demonstrate the average movecount of each mother, nor are these the most efficient solutions. These are just to demonstrate the steps of the method.

 

[Cubing Forever]

Here's an idea(OH-only) which I was thinking about for a few weeks:

Spoiler: Motivation

CP(or 2-generator reduction) is a cool concept, but I feel that most methods that involve it use it in the wrong place, i.e the start of the solve. This would result in a waste of inspection, the inability to be colour neutral and lack of choices. This Mehta variant that I'm proposing has CP later in the solve which allows you to be colour neutral.

Spoiler: Steps

Step 1: Mehta-style FB
Step 2: Belt
Step 3: DCAL(80 cases)
Step 4: CPEO(25 cases excluding the cases which are AUFs to other cases and the cases where the edges are oriented)
Step 5(and 6): one of two variants below

Spoiler: Variants

Here are the variants for steps 6 and 7:
Variant 1: Solve the rest of the cube in 1 alg(2GLL+1)(408 cases)
Variant 2: CDRLL(or OCLL since there's only 7 cases)+L5EP
Variant 3: JTLE+EPLL

Spoiler: Movecounts(FB+belt+DCAL+CPEO)

ok, let's be honest here(for full CN):

FB: 6-6.5
Belt: 9-10
DCAL: 9-10
CPEO: 10-12
Total: 34-38

Spoiler: Movecounts(finish)

2GLL+1: 14-14.5 according to @Athefre iirc
CDRLL+L5EP: 9.5+10=19.5
JTLE+EPLL: 10+10=20

Total avg movecount= 48-54 moves

Spoiler: Pros

Allows for a 2-gen finish while allowing full CN(which most other CP methods don't allow).
Has a movecount comparable to Roux/ZZ-a.
Could be a viable alternative to Roux/ZZ-a if CPEO recog is fixed.

Spoiler: Cons

High algcount
FB+belt ergonomics can be bad for some people.
I can't seem to come up with a good recog system for CPEO(If someone comes up with an idea I'd be super grateful to them).
There isn't any algsheet for 2GLL+1 yet afaik.

Spoiler: Example solves

R2 U2 R D2 R2 B2 R B2 U2 R2 U2 F R' U' F' D L' U' F' D'

2GLL+1 version: 46 STM
D2 R F L R U2 R' //FB
D2 R2 u' R' u2 R //belt
R2 U' R2 U R2 U' R2 //DCAL
u U R' U2 R U2 F R F' //CPEO
R U R' U R U' R' U R' U' R U' R' U R' U R //2GLL+1

CDRLL+L5EP version: 54 STM
D2 R F L R U2 R' //FB
D2 R2 u' R' u2 R //3QB
R2 U' R2 U R2 U' R2 //DCAL
u U R' U2 R U2 F R F' //CPEO
U R U2 R' U' R U' R2 U2 R U R' U R //CDRLL
U R' U' R U R U R U' R' U2 //L5EP

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

2GLL+1 version: 48 STM
x y //inspection
F' E U F' r2 U2 r2 //FB
R' E R' U' R u' f' U f //Belt
U R U' R2 U R2 U' R //DCAL
U2 r U' r' U' r U r' F' U F //CPEO
U2 R U2 R2 U' R2 U' R' U R' U R U2 //2GLL(+1)

JTLE+EPLL version: 54 STM(53 w/cancellation)
x y //inspection
F' E U F' r2 U2 r2 //FB
R' E R' U' R u' f' U f //Belt
U R U' R2 U R2 U' R //DCAL
U2 r U' r' U' r U r' F' U F //CPEO
U2 R U2 R' U' R U' R' //JTLE
R2 U' R' U' R U R U R U' R //EPLL

I'd name it Mehta-CPEO or something ig.
Anyone is welcome to voice their opinions/concerns for this.
Thank you.

 

[V Achyuthan]

Here's an idea(OH-only) which I was thinking about for a few weeks:

Spoiler: Motivation

CP(or 2-generator reduction) is a cool concept, but I feel that most methods that involve it use it in the wrong place, i.e the start of the solve. This would result in a waste of inspection, the inability to be colour neutral and lack of choices. This Mehta variant that I'm proposing has CP later in the solve which allows you to be colour neutral.

Spoiler: Steps

Step 1: Mehta-style FB
Step 2: Belt
Step 3: DCAL(80 cases)
Step 4: CPEO(25 cases excluding the cases which are AUFs to other cases and the cases where the edges are oriented)
Step 5(and 6): one of two variants below

Spoiler: Variants

Here are the variants for steps 6 and 7:
Variant 1: Solve the rest of the cube in 1 alg(2GLL+1)(408 cases)
Variant 2: CDRLL(or OCLL since there's only 7 cases)+L5EP
Variant 3: JTLE+EPLL

Spoiler: Movecounts(FB+belt+DCAL+CPEO)

ok, let's be honest here(for full CN):

FB: 6-6.5
Belt: 9-10
DCAL: 9-10
CPEO: 10-12
Total: 34-38

Spoiler: Movecounts(finish)

2GLL+1: 14-14.5 according to @Athefre iirc
CDRLL+L5EP: 9.5+10=19.5
JTLE+EPLL: 10+10=20

Total avg movecount= 48-53 moves

Spoiler: Pros

Allows for a 2-gen finish while allowing full CN(which most other CP methods don't allow).
Has a movecount comparable to Roux/ZZ-a.
Could be a viable alternative to Roux/ZZ-a if CPEO recog is fixed.

Spoiler: Cons

High algcount
FB+belt ergonomics can be bad for some people.
I can't seem to come up with a good recog system for CPEO(If someone comes up with an idea I'd be super grateful to them).
There isn't any algsheet for 2GLL+1 yet afaik.

I'd name it Mehta-CPEO or something ig.
Anyone is welcome to voice their opinions/concerns for this.
Thank you.

This looks cool. I might try it once the algs are finished. Also how to gen algs for CP?

 

[Cubing Forever]

This looks cool. I might try it once the algs are finished. Also how to gen algs for CP?

Cool. CPEO algs are finished though. Just a few optimizations are required here and there. It's the recog method that isn't finished yet.

 

[V Achyuthan]

I found a new method (sort of).

Spoiler: Steps

1. EO + Build 1x2x3 block like the Roux method except it can have any U/D colour on the bottom side and any E slice edge on the E slice.
2. do the same on the other side except when inserting the last pair use WV to Orient corners
Finish any ways you want to

This is just a fun method and not really a speedsolving method.

Spoiler: Example solve 1 45 STM

F2 R2 D B2 L2 B2 U' B2 F2 L2 B' R D2 R' U2 R' B2 D' F D
F' U2 L D B // EO
D' F2 D // FB
U R' U2 R U' M2 U2 R U' R' // SB
L' U' L U' L' U2 L // CO
B2 // corners
x2 S U2 S' M2 U r2 U2 R2 U2 R2 y // DB,DL and DR + parity
U2 M U2 M U M' U2 M' U2 // L5EP

Spoiler: Example solve 2 50 STM

y' B2 D2 B // EO
D' R' F2 D // FB
U2 R U R2 U' r2 U2 U R' U' R2 U' R2 U2 R // SB + CO
U2 R2 U2 B2 R2 U' F2 U R2 U R2 U' F2 // F2L-1
y' U' R2 U R2 U' R2 D R2 U' R2 U R2 U' D' U' // TTLL

Let me know your thoughts on this. Can this be used for speedsolving?

 

[V Achyuthan]

Cool. CPEO algs are finished though. Just a few optimizations are required here and there. It's the recog method that isn't finished yet.

ya I understand. Just asked how to gen CP algs. Also can you add some example solves to the main post?

 

[Cubing Forever]

ya I understand. Just asked how to gen CP algs. Also can you add some example solves to the main post?

You can watch this to know how to gen CP algs. I've got two example solves in the 3x3 example solve thread which I'll add later.

 

[V Achyuthan]

You can watch this to know how to gen CP algs. I've got two example solves in the 3x3 example solve thread which I'll add later.

ok thanks

 

[OreKehStrah]

Here's an idea(OH-only) which I was thinking about for a few weeks:

Spoiler: Motivation

CP(or 2-generator reduction) is a cool concept, but I feel that most methods that involve it use it in the wrong place, i.e the start of the solve. This would result in a waste of inspection, the inability to be colour neutral and lack of choices. This Mehta variant that I'm proposing has CP later in the solve which allows you to be colour neutral.

Spoiler: Steps

Step 1: Mehta-style FB
Step 2: Belt
Step 3: DCAL(80 cases)
Step 4: CPEO(25 cases excluding the cases which are AUFs to other cases and the cases where the edges are oriented)
Step 5(and 6): one of two variants below

Spoiler: Variants

Here are the variants for steps 6 and 7:
Variant 1: Solve the rest of the cube in 1 alg(2GLL+1)(408 cases)
Variant 2: CDRLL(or OCLL since there's only 7 cases)+L5EP
Variant 3: JTLE+EPLL

Spoiler: Movecounts(FB+belt+DCAL+CPEO)

ok, let's be honest here(for full CN):

FB: 6-6.5
Belt: 9-10
DCAL: 9-10
CPEO: 10-12
Total: 34-38

Spoiler: Movecounts(finish)

2GLL+1: 14-14.5 according to @Athefre iirc
CDRLL+L5EP: 9.5+10=19.5
JTLE+EPLL: 10+10=20

Total avg movecount= 48-54 moves

Spoiler: Pros

Allows for a 2-gen finish while allowing full CN(which most other CP methods don't allow).
Has a movecount comparable to Roux/ZZ-a.
Could be a viable alternative to Roux/ZZ-a if CPEO recog is fixed.

Spoiler: Cons

High algcount
FB+belt ergonomics can be bad for some people.
I can't seem to come up with a good recog system for CPEO(If someone comes up with an idea I'd be super grateful to them).
There isn't any algsheet for 2GLL+1 yet afaik.

Spoiler: Example solves

2GLL+1 version: 46 STM
D2 R F L R U2 R' //FB
D2 R2 u' R' u2 R //belt
R2 U' R2 U R2 U' R2 //DCAL
u U R' U2 R U2 F R F' //CPEO
R U R' U R U' R' U R' U' R U' R' U R' U R //2GLL+1

CDRLL+L5EP version: 54 STM
D2 R F L R U2 R' //FB
D2 R2 u' R' u2 R //3QB
R2 U' R2 U R2 U' R2 //DCAL
u U R' U2 R U2 F R F' //CPEO
U R U2 R' U' R U' R2 U2 R U R' U R //CDRLL
U R' U' R U R U R U' R' U2 //L5EP

I'd name it Mehta-CPEO or something ig.
Anyone is welcome to voice their opinions/concerns for this.
Thank you.

for CPEO Recog, just split it into recognizing the COLL case to determine the CP, then the EO case

 

[Cubing Forever]

for CPEO Recog, just split it into recognizing the COLL case to determine the CP, then the EO case

Yeah I did think about this but I feel that the information might be insufficient to determine which two corners have to be swapped.

 

[OreKehStrah]

Yeah I did think about this but I feel that the information might be inefficient to determine which two corners have to be swapped.

You just need to be able to recognize each COLL, and then have memorized what COLL case for each OCLL corresponds to each CP. it really shouldn’t be an issue

 

[Cubing Forever]

You just need to be able to recognize each COLL, and then have memorized what COLL case for each OCLL corresponds to each CP. it really shouldn’t be an issue

Thanks.
EDIT: idk why but this actually reminds me of ROLL.

 

[V Achyuthan]

I thought of a LSLL thing
before solving the Last slot build a 1x1x3 block and solve the LS while preserving the 1x1x3 block. And then 1LLL. This reduces the number of 1LLLs that need to be learned.
Here are some examples

Spoiler: Example solve 1 53 STM

R2 B R2 U2 B2 F U2 F L2 F D2 U2 R' B D2 L U B2 R2 D' R'
R U R U' D' R' U R' F D U F' L' U' L D' // xx-cross
U' F U' F' // 3rd pair
U2 R U' R' U F' U F // 113 Block
R' D' R U2 R' D R U R U' R' // 4th pair while preserving 113 block
U' M' U R U' r' U R U R' U F' U F // 1LLL

Spoiler: Example solve 2 53 STM

F' R' L2 D R' U2 F L2 F R' B2 U2 B2 U2 R D2 L' U2 R' F2 U2
z D' F2 D2 F' // Cross
U2 R' L U2 L' R // 1st pair
U R2 u R2' u' R2 // 2nd pair
y2 R' U2 R U R' U' R // 3rd pair
U F' U F // 113 Block
U2 R U R' U' R' D' R U2 R' D R // 4th pair while preserving 113 block
R' U2 R U R' U R f R U R' U' f' U // 1LLL

 

[Thom S.]

I thought of a LSLL thing
before solving the Last slot build a 1x1x3 block and solve the LS while preserving the 1x1x3 block. And then 1LLL. This reduces the number of 1LLLs that need to be learned.

Thought of. Called Line and Anti-Line subsets

 

[V Achyuthan]

Thought of. Called Line and Anti-Line subsets

lol. ok then I had thought of another LSLL as well
Solve the last slot while permuting the edges (yes without orienting them).
Here are 2 examples

1. R2 U' B2 R2 F2 R2 F2 D' F2 D' L2 U R' B' D2 U2 B2 L D F2 R2 B2
x U L U' L F' // Cross
D' L' U2 L u // 1st pair
U2 L F' U F L' // 2nd pair
R' U R2 U R' // 3rd pair (5)
y R' U2 R2 U R2 U' R U R' U R // 4th pair + Permute edges
U' R' U2 R2 U R2 U R U' R U' R' U // ZBLL/1LLL

2. D2 R2 B2 D U2 F2 L2 U' F2 R2 U2 B2 R' B' D' B R U2 F' L' D
F2 R2 B' D R U' R' L' // xcross
y R U2' R2' U' R2 U R' L U' L' // 2nd & 3rd pairs
U2 R' F R F' R U' R' U2 R U2 R' // 4th pair + Permute edges
y' r U R' U R U2 r' U R U2 R' U' R U' R' U // 1LLL

 

[PapaSmurf]

That gives you some just straight up bad 1LLLs. You might as well just do ZB. Just generally, don't use this thread as a method idea dump. Instead, try to think through the method properly and, if it came to you in about 5 mins, why no one else has mentioned it.

 

[V Achyuthan]

That gives you some just straight up bad 1LLLs. You might as well just do ZB. Just generally, don't use this thread as a method idea dump. Instead, try to think through the method properly and, if it came to you in about 5 mins, why no one else has mentioned it.

OK ig