The Orcuber's Hunt For The Best 3x3 Method

[Jared Burkhart]

Okay so far the team (Or people who have shown interest in working on this) include @CrossPetrus @speedsolvetheory @-Cube Kid- @Martyt7777 @Michael Wimberly @Caleb Valenzuela @Jared Burkhart.
This leads me to my next question what do we want to work on I have some ideas if you have any other ideas I would love to hear them. We could do:
Zbll optimization
APB development
CFOP Variation
New Method
(Just some ideas that were mentioned)

I think we should start with APB optimization.

I focus primarily on xCross, cross +1 and last layer optimization, so see you guys later!

 

[Martyt7777]

Okay so far the team (Or people who have shown interest in working on this) include @CrossPetrus @speedsolvetheory @-Cube Kid- @Martyt7777 @Michael Wimberly @Caleb Valenzuela @Jared Burkhart.
This leads me to my next question what do we want to work on I have some ideas if you have any other ideas I would love to hear them. We could do:
Zbll optimization
APB development
CFOP Variation
New Method
(Just some ideas that were mentioned)

I have a ZB variation idea, can I share it?

 

[acrystal]

I can join, I main many methods I made.

 

[speedsolvetheory]

Really ZB is as optimal as you can get. Speedcubing isn't about thinking. The first step, is basically an alg you make up in 8-12s, then you do 4 mini algs, and 2 sets of longer algs. So basically you are just doing algs the whole solve. EO is slow because it has bad ergonomics, not enough to justify taking time to do EOline or Eocross. Really CFOP and ZB are the optimal methods, because they are essentially just repeated alogirthms across the entire solve. This principle was recently challenged by Teodor's freefop solution, but that was a freak occurence. We should focus on f2l and zbls and zbll optimization. Every method every conceived has an inutitive segment. F2L can be interpreted as intuitive but it is not. APB Mehta are trash
New algs for awkward shape: M U F R U R' U' F' M'
r R2 F' U' F U R U' M
I know they aren't the best but I like the first one

 

[Sub20Soon]

I could join and i will try to document the method and do a guide about it. I could also check the average movecount and brainstrom ideas.

 

[speedsolvetheory]

Count me out actualy since I basically explained why I think ZB is the optimal method

 

[Martyt7777]

I have a method idea that is a variation of ZB, not sure if 1296 is too many algs though

Spoiler: ZB-K method

ZB-K (placeholder name, might change in the future) is a method that forces a ZBLL that is not H or Pi. It has ~1296 algorithms.

Steps​

Cross+3​

The usual.

ZBKLS​

ZBKLS solves LS and EO while guaranteeing at least 1 oriented corner. It is recognised like ZBLS, but you have to recognise a yellow corner. This has ~915 algs.

Based off of the orientation of the yellow corner, you have to know which of the 3 ZBLS algs you should do (ZBKLS) to orient that corner. To know which yellow corner to recognise, follow this guide:

If the white corner is not in the slot, the DFR corner is reocgnised; Otherwise, the URF corner. For cases where both pieces of the last pair are misoriented in the slot, you can choose any yellow corner.

Because the insertion of the 3rd pair usually does not affect the pieces in the last pair slot (and the majority of last pair cases have the last corner outside of the slot), you can easily lookahead to the orientation of the yellow corner.

Examples for the DFR corner (pay attention to the corner when recognising):

Spoiler: DFR facing D

 Twizzle link 

 Setup 

z2 R U R' U R U' R' U'

 Moves 

U R U R' U' R U' R'

Spoiler: DFR facing F

 Twizzle link 

 Setup 

z2 R U2 R' U2

 Moves 

U2 R U2' R'

Spoiler: DFR facing R

 Twizzle link 

 Setup 

z2 R U R' U'

 Moves 

U R U' R'

Examples for the URF corner (pay attention to the corner when recognising):

Spoiler: URF facing U

 Twizzle link 

 Setup 

z2 R' U' R' U' R U R U R

 Moves 

R' U' R' U' R' U R U R

Spoiler: URF facing R

 Twizzle link 

 Setup 

z2 R U R' U2 R U R' U2 R U' R' U

 Moves 

U' R U R' U2' R U' R' U2' R U' R'

Spoiler: URF facing F

 Twizzle link 

 Setup 

z2 F' R U R' U' R' F R

 Moves 

R' F' R U R U' R' F

ZBLL (I don’t think I’ll call this ZBKLL)​

ZBLL here finishes off the cube. But it’s slightly different in the ZB-K method. As ZBKLS oriented a corner, H and Pi aren’t required. This has ~381 algs. (including PLL)

Pros and cons​

Pros​

• H and Pi (the worst ZBLL sets to recog + execute) are gone
• Fast sets such as S, aS, and O (PLL) are more likely to appear

Cons​

• Very high alg count
• Some ZBKLS algs (in theory) are inefficient
• ZBKLS recog is slightly harder than ZBLS recog

 

[OreKehStrah]

I have a method idea that is a variation of ZB, not sure if 1296 is too many algs though

Spoiler: ZB-K method

ZB-K (placeholder name, might change in the future) is a method that forces a ZBLL that is not H or Pi. It has ~1296 algorithms.

Steps​

Cross+3​

The usual.

ZBKLS​

ZBKLS solves LS and EO while guaranteeing at least 1 oriented corner. It is recognised like ZBLS, but you have to recognise a yellow corner. This has ~915 algs.

Based off of the orientation of the yellow corner, you have to know which of the 3 ZBLS algs you should do (ZBKLS) to orient that corner. To know which yellow corner to recognise, follow this guide:

If the white corner is not in the slot, the DFR corner is reocgnised; Otherwise, the URF corner. For cases where both pieces of the last pair are misoriented in the slot, you can choose any yellow corner.

Because the insertion of the 3rd pair usually does not affect the pieces in the last pair slot (and the majority of last pair cases have the last corner outside of the slot), you can easily lookahead to the orientation of the yellow corner.

Examples for the DFR corner (pay attention to the corner when recognising):

Spoiler: DFR facing D

 Twizzle link 

 Setup 

z2 R U R' U R U' R' U'

 Moves 

U R U R' U' R U' R'

Spoiler: DFR facing F

 Twizzle link 

 Setup 

z2 R U2 R' U2

 Moves 

U2 R U2' R'

Spoiler: DFR facing R

 Twizzle link 

 Setup 

z2 R U R' U'

 Moves 

U R U' R'

Examples for the URF corner (pay attention to the corner when recognising):

Spoiler: URF facing U

 Twizzle link 

 Setup 

z2 R' U' R' U' R U R U R

 Moves 

R' U' R' U' R' U R U R

Spoiler: URF facing R

 Twizzle link 

 Setup 

z2 R U R' U2 R U R' U2 R U' R' U

 Moves 

U' R U R' U2' R U' R' U2' R U' R'

Spoiler: URF facing F

 Twizzle link 

 Setup 

z2 F' R U R' U' R' F R

 Moves 

R' F' R U R U' R' F

ZBLL (I don’t think I’ll call this ZBKLL)​

ZBLL here finishes off the cube. But it’s slightly different in the ZB-K method. As ZBKLS oriented a corner, H and Pi aren’t required. This has ~381 algs. (including PLL)

Pros and cons​

Pros​

• H and Pi (the worst ZBLL sets to recog + execute) are gone
• Fast sets such as S, aS, and O (PLL) are more likely to appear

Cons​

• Very high alg count
• Some ZBKLS algs (in theory) are inefficient
• ZBKLS recog is slightly harder than ZBLS recog

That massively overcomplicates things. You really only need a list of speed optimal ZBLS algs, then document the 6/27 CO cases that lead to H/Pi and have an alternate ZBLS alg.

It's not really worth further complicating your LS just to avoid HP imo, particularly when ZBLS is still developing and being optimized.

 

[0rcoraCuber21]

I have a method idea that is a variation of ZB, not sure if 1296 is too many

I would say it is but like thats what everyone said about 800 algs for zb and now we have 5 year olds learning it. But this method does seem a little complex.

 

[Caleb Valenzuela]

APB Mehta are trash

Absolutely not. Mehta might be trash, but APB is NOT. It’s a very alg heavy method as stated by its developers and several users of the method. BR pair uses algs, EO uses algs, LXS uses algs, and then comes ZB for the LL step, which i don’t really have to explain. This all tallies to around 700 algorithms. Stating that APB is trash contradicts your earlier statement:

Really CFOP and ZB are the optimal methods, because they are essentially just repeated algorithms across the entire solve

Sure, 223 is usually intuitive, but cross in CFOP and ZB are intuitive too, same goes for xcrosses. Also 1x2x3 can be easily planned in inspection and then you can just lookahead to 223. Some people even plan full 223 in inspection.

Speedcubing isn't about thinking

Then explain how people can predict full F2L.

 

[0rcoraCuber21]

Absolutely not. Mehta might be trash, but APB is NOT. It’s a very alg heavy method as stated by its developers and several users of the method. BR pair uses algs, EO uses algs, LXS uses algs, and then comes ZB for the LL step, which i don’t really have to explain. This all tallies to around 700 algorithms. Stating that APB is trash contradicts your earlier statement:

Sure, 223 is usually intuitive, but cross in CFOP and ZB are intuitive too, same goes for xcrosses. Also 1x2x3 can be easily planned in inspection and then you can just lookahead to 223. Some people even plan full 223 in inspection.

Then explain how people can predict full F2L.

I feel like you really like APB and want to work on finding new Optimizations like zb for cfop. (I would be down to learn and work on this)

 

[acrystal]

What about Lau?
(Inverted cross)-1 -> Inverted 1x2x3 -> Fix the 2 edges opposite to the corners (D2 to get 2x2x3 block) -> Fix last white edge -> insert F2L pair -> VLS -> PLL

 

[AJT17]

What about Lau?
(Inverted cross)-1 -> Inverted 1x2x3 -> Fix the 2 edges opposite to the corners (D2 to get 2x2x3 block) -> Fix last white edge -> insert F2L pair -> VLS -> PLL

This is just a much more inefficient way of creating a 2x2x3 block, and it doesn't add anything unique. It's just an all-around worse version of Petrus, since you don't even do EO after 223, which just makes it worse.

 

[Caleb Valenzuela]

I feel like you really like APB and want to work on finding new Optimizations like zb for cfop. (I would be down to learn and work on this)

I hold a WB with the method so of course i have to defend it
I’m confident @AJT17 will agree as well

 

[Martyt7777]

What about Lau?
(Inverted cross)-1 -> Inverted 1x2x3 -> Fix the 2 edges opposite to the corners (D2 to get 2x2x3 block) -> Fix last white edge -> insert F2L pair -> VLS -> PLL

We all saw that video, didn’t we?

I feel like the 2x2x3 block is too systematic. Lau would be better with blockbuilding. It is an interesting method though

 

[Caleb Valenzuela]

We all saw that video, didn’t we?

I feel like the 2x2x3 block is too systematic. Lau would be better with blockbuilding. It is an interesting method though

uhh
there are other ways to make a 2x2x3 block
such as block building
it’s just that 1x2x3 into DFDB is the most popular

 

[Jared Burkhart]

uhh
there are other ways to make a 2x2x3 block
such as block building
it’s just that 1x2x3 into DFDB is the most popular

Well, I think that it is worth trying. But aren't we working on making a new approach? he idea of this thread was to develop an entirely new method, but the problem is that whatever approach we think of is either already developed or heavily based off of a ton of other techniques...this is challenging for me

 

[acrystal]

Well, I think that it is worth trying. But aren't we working on making a new approach? he idea of this thread was to develop an entirely new method, but the problem is that whatever approach we think of is either already developed or heavily based off of a ton of other techniques...this is challenging for me

Yes, you are right. But we are taking methods we are fond of and we will make a beautiful method out of this, just wait, shall we?

 

[Jared Burkhart]

Yes, you are right. But we are taking methods we are fond of and we will make a beautiful method out of this, just wait, shall we?

indeed. I love xCross

 

[Martyt7777]

Just an idea… but what about solving LXS and EO at the same time for APB?