The New Method / Substep / Concept Idea Thread

[Teoidus]

Serious suggestion?

 

[mDiPalma]

why dont u do Y perm, E perm, Ua perm for every single move in the PLL?

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

and then you can do Y perm, E perm, Ua perm for every single move of the Y perm, E perm, Ua perm

 

[crafto22]

Don't have much time to elaborate on this idea atm, I'll be back soon to explain some more but basically I was bored during a lecture and came up with this:
1. FB (8/8)
2. Belt + CO (9/17)
3. CP, details coming soon, 18 algs (6/23)
4. DR edge + influence L6E (4/27)
5. L6E (12/39)

 

[Neuro]

Wait how exactly are you influencing L6E? It seem like L6E should take more moves than that.

 

[Teoidus]

Probably getting a favorable EOLR case.

Though, I think his movecounts are a little optimistic

 

[crafto22]

Probably getting a favorable EOLR case.

Though, I think his movecounts are a little optimistic

Ya I think I'm pushing it a little when it comes to move count. Idk tho, a lot of the CP cases would be as little as 3 moves... Any other parts that may be unrealistic?

 

[Hazel]

An idea I came up with (probably really bad):
1) Solve Roux 1x2x3 block on left
2) Solve remaining 6 corners
3) L9E (many cases)

You could also decrease the amount of L9E algorithms/cases to an extent by solving any D layer edges that can be solved with just M and U moves in between steps 2 and 3 (because total L9E cases - all L9E cases where there's an unsolved D layer edge that can be solved with M and U moves = a slightly smaller number of L9E cases).

Pros:
- Uses 3 looks total at least, more if you can't one-look the Roux block, the corners, and L9E
- According to my one example solve, it doesn't use too many moves if you're really efficient

Cons:
- Quite a bit of L9E cases
- taking advantage of the pros of this method requires a lot of skill

Example solve:
Scramble: F2 L B2 L R' B R' B2 F U L' F2 D2 L R2 F U' B R2

1) z2 U R' B2 U2 F D F D2 (8/37)
2) U F R' F2 U' F R U2 R2 U (10/37)
2.5) M' U2 M U2 (4/37)
3) y R F U F2 U D R' D' F' U' F U' F U R' (15/37)

 

[Thermex]

An idea I came up with (probably really bad):
1) Solve Roux 1x2x3 block on left
2) Solve remaining 6 corners
3) L9E (many cases)

You could also decrease the amount of L9E algorithms/cases to an extent by solving any D layer edges that can be solved with just M and U moves in between steps 2 and 3 (because total L9E cases - all L9E cases where there's an unsolved D layer edge that can be solved with M and U moves = a slightly smaller number of L9E cases).

Pros:
- Uses 3 looks total at least, more if you can't one-look the Roux block, the corners, and L9E
- According to my one example solve, it doesn't use too many moves if you're really efficient

Cons:
- Quite a bit of L9E cases
- taking advantage of the pros of this method requires a lot of skill

Example solve:
Scramble: F2 L B2 L R' B R' B2 F U L' F2 D2 L R2 F U' B R2

1) z2 U R' B2 U2 F D F D2 (8/37)
2) U F R' F2 U' F R U2 R2 U (10/37)
2.5) M' U2 M U2 (4/37)
3) y R F U F2 U D R' D' F' U' F U' F U R' (15/37)

@Aerma This is EXACTLY what everyone here was been working on for the past couple of weeks. It's been determined for L6C the fastest way is TEG (Twisty EG) but we're still working on ways of solving L9E. (Btw in case you didn't know this method is called WaterRoux).

 

[Hazel]

Oh sorry, I thought WaterRoux was something different!
Every time I come up with some method idea, it's either taken, really bad, or not original enough :/

 

[Neuro]

I'm working on the DF+EO algs right now, if anyone want to help out PM me your email. Here's the sheet, note that DF is in the FR slot. I'd like to keep all the algs using <R,r,U,> if possible.

 

[Hazel]

Is this new?:
1) Solve Petrus block + 1x1x3 bar connected to it
2) Solve remaining 4 corners
3) L6E

Kinda similar to WaterRoux, but instead of a Roux block it's a Petrus block (+ a bit more), meaning a longer first step but only L6E instead of L9E/L7E, and a lot less corner cases with this, too, only 42!

I was also thinking of a variant where it's a Petrus block + ONE edge instead of a whole 1x1x3 block, so the first step wouldn't take quite as long, but you need to solve 6 corners as opposed to 4.

 

[Neuro]

Oh sorry, I thought WaterRoux was something different!
Every time I come up with some method idea, it's either taken, really bad, or not original enough :/

Hey being the creator of WaterRoux, NLS, and one of the creators of NS4, I can honestly say that good ideas are fairly rare even for the best method designers. All of us have had our duds, my first proposals were absolutely awful (ZZ with UL/UR in the line to do LSE and an awful WVCP-based method) and most ideas don't get very far. The key is to always keep trying. If you push through all of the failed attempts, you'll find what works and what doesn't, and be able to use that knowledge to create something great. It's all about tenacity and learning from your mistakes. So don't worry, you'll probably come up with something good in time!

EDIT: Aerma, here are the official steps to WaterRoux

1-FB
2-Corners
3-ERL: Solve 2 or more R layer edges (Edged of Right Layer)
4-LxE: The highest possible amount of edges is 7, but L6 and L5 edges is common

There seems to be a misconception that it's L9E. It kinda is, but it's in 2 sections and is not truly L9E.

 

[crafto22]

Another method I thought might be good:
1. FB (8/8)
2. Orient the remaining E-slice edges and orient the edges at DF and DB (4/12)
3. Solve the E-slice and orient the rest of the D-layer (9/21)
4. OLL, lower move count due to not having to preserve F2L (7/28)
5. CP (6/34)
6. L7EP (12/46)

Easy recog, more algs tho (57 for OLL + 18 for CP)

 

[Teoidus]

Ya I think I'm pushing it a little when it comes to move count. Idk tho, a lot of the CP cases would be as little as 3 moves... Any other parts that may be unrealistic?

I'm skeptical of the CP, belt+CO, and DR+influence estimates

Though, I could still see it getting movecounts comparable to roux

 

[crafto22]

I'm skeptical of the CP, belt+CO, and DR+influence estimates

Though, I could still see it getting movecounts comparable to roux

Ya I tried generating algs and I was totally off in terms of CP. The algs are bad and so is recog, I'm scrapping the whole idea.

 

[Hazel]

So I had the idea of trying to find a bunch of Last Layer cases that could be solved with one look by using 2 easy OLL cases instead of OLL then PLL, and have easy recognition. I think Kirjava proposed something like this before, but I don't remember it fully; I think his version didn't have any algorithms or something? Anyway, I found 2 already, if anyone wants to comment on them.

 

[crafto22]

Hey guys I'm trying to find a good way of implementing Guimond (2x2 method) into something for 3x3. Any suggestions?

 

[crafto22]

Something like this possibly:
1. Three E-slice edges and three oriented corners on D (8 moves)
2. CO (around 4 moves, according to the Guimond wiki page)
3. Corner separation (3 moves average according to the Guimond wiki)
4. CP + DL edge (8.5 moves)
5. L7E (17 moves)
Total: 39.5 moves on average. Did I get anything wrong?

 

[crafto22]

Sorry I did get something wrong, it's an L8E finish, probably closer to 42 moves total. Still good I guess. Not the sub-40 I was looking for.

 

[Teoidus]

Something like this possibly:
1. Three E-slice edges and three oriented corners on D (8 moves)
2. CO (around 4 moves, according to the Guimond wiki page)
3. Corner separation (3 moves average according to the Guimond wiki)
4. CP + DL edge (8.5 moves)
5. L7E (17 moves)
Total: 39.5 moves on average. Did I get anything wrong?

Hm, I'm not sure if corner separation in 3 moves is possible while preserving the E slice.

Also, having generated random CP algs in the past I'm tempted to say CP + DL will be more around 10-11 moves. That's just a guess though.